Merge branch 'release/1.13.0' into main

.gitignore

@@ -1,4 +1,12 @@
 # VS Code — only shared workspace settings
 .vscode/*
+!.vscode/
 !.vscode/settings.json
 !.vscode/extensions.json
+
+# framework
+/master-knowledge-genome/
+/keys/
+*.key
+__pycache__/
+*.pyc

Makefile

@@ -1,5 +1,5 @@
 # =============================================================================
-# Knowledge Genome - Makefile v. 1.1.6
+# Knowledge Genome - Makefile v. 1.13.0
 # Orchestrates the setup and management of the knowledge base.
 # =============================================================================
 
@@ -11,7 +11,7 @@ export $(shell grep -v '^[#[:space:]]' globals.env | sed 's/=.*//')
 help:
 	@echo "Available commands:"
 	@echo "  make setup            - Full system initialization"
-	@echo "  make add-genome       - Register and scaffold a new genome [LINKED=owner/repo]"
+	@echo "  make add-genome       - Register and scaffold a new genome [LINKED=owner/repo] [CROSS=yes|no]"
 	@echo "  make status           - Check submodule and encryption status"
 	@echo "  make lint             - Verify schema, privacy flags, and metadata"
 	@echo "  make verify-structure - Report directory drift across all genomes"
@@ -30,16 +30,17 @@ setup:
 add-genome:
 	@if [ -z "$(NAME)" ] || [ -z "$(DESC)" ]; then \
 		echo "Error: NAME and DESC are required."; \
-		echo "Usage: make add-genome NAME=my-genome DESC='My description' [LINKED=owner/project-repo]"; \
+		echo "Usage: make add-genome NAME=my-genome DESC='My description' [LINKED=owner/project-repo] [CROSS=yes|no]"; \
 		exit 1; \
 	fi
-	@bash scripts/add-genome.sh "$(NAME)" "$(DESC)" "$(LINKED)"
+	@bash scripts/add-genome.sh "$(NAME)" "$(DESC)" "$(LINKED)" "$(or $(CROSS),no)"
 
 status:
+	@[ -d "$(MASTER_REPO)" ] || { echo "Master non trovato. Esegui 'make setup'."; exit 1; }
 	@echo "--- Master Status ---"
-	@git submodule status
+	@cd $(MASTER_REPO) && git submodule status
 	@echo "--- Encryption Status (per genome) ---"
-	@git submodule foreach 'git-crypt status 2>/dev/null | head -n 10 || true'
+	@cd $(MASTER_REPO) && git submodule foreach 'git-crypt status 2>/dev/null | head -n 10 || true'
 
 verify-structure:
 	@bash scripts/verify-genomes.sh
@@ -62,14 +63,16 @@ doctor:
 	@echo "System ready."
 
 sync:
+	@[ -d "$(MASTER_REPO)" ] || { echo "Master non trovato. Esegui 'make setup'."; exit 1; }
 	@echo "Syncing submodules..."
-	@git submodule update --init --recursive
+	@cd $(MASTER_REPO) && git submodule update --init --recursive
 	@echo "--- Unpushed commits per genome ---"
-	@git submodule foreach 'git log --oneline @{u}.. 2>/dev/null | head -5 || true'
+	@cd $(MASTER_REPO) && git submodule foreach 'git log --oneline @{u}.. 2>/dev/null | head -5 || true'
 
 lock:
+	@[ -d "$(MASTER_REPO)" ] || { echo "Master non trovato. Esegui 'make setup'."; exit 1; }
 	@echo "Locking master repository..."
-	@git-crypt lock 2>/dev/null || true
+	@cd $(MASTER_REPO) && git-crypt lock 2>/dev/null || true
 	@echo "Locking all submodules..."
-	@git submodule foreach 'git-crypt lock 2>/dev/null || true'
+	@cd $(MASTER_REPO) && git submodule foreach 'git-crypt lock 2>/dev/null || true'
 	@echo "All genomes securely locked."

README.md

@@ -77,6 +77,11 @@ master-knowledge-genome/              ← Root orchestrator (submodule registry)
 └── AGENTS.md                         ← Global coordination schema (cross-genome rules)
 ```
 
+> The genome names above (`genome-dev`, `genome-finance`, `genome-homelab`) are
+> **illustrative** — they show the kind of multi-domain layout this orchestrator targets.
+> The shipped `registry.sh` defines a single disposable sandbox, **`genome-test`**; you
+> create real genomes yourself with `make add-genome` (see the registry examples below).
+
 Each genome is an independent git repository:
 
 ```text
@@ -175,6 +180,11 @@ knowledge-genome-orchestrator/        ← This repository (setup tooling)
 > The `skills/ingest/` directory is version-controlled here but **deployed** to the AI
 > node (vm101) under `~/.pi/agent/skills/ingest`. The agent (`pi`) does only semantic work
 > and writes a manifest; `run-ingest.sh` does the mechanical steps. See [Workflows → Ingest](#ingest).
+>
+> ingest-semantic.py: one schema-constrained call to local model, returns JSON. run-ingest.sh: index/log/lint/PR.
+> Semantic JSON extraction → deterministic wiki conform + manifest.
+>
+> cp skills/ingest/\* ~/.pi/agent/skills/ingest/ after make setup. Updated via git pull on laptop, pushed to vm101 via SSH in n8n flow.
 
 ---
 
@@ -330,13 +340,14 @@ WORK_DIR="${HOME}/knowledge-genome-orchestrator"
 KEYS_DIR="${WORK_DIR}/keys"
 
 # Genome registry — format: "name|description|linked_repo"
-# The third field is OPTIONAL:
+# The third and fourth fields are OPTIONAL:
 #   - leave it empty  → knowledge-only genome (no linked project)
 #   - owner/repo      → genome is linked to that project repository (rendered into AGENTS.md)
+#   - cross_source    → yes|no (default no): whether the cross-genome collector may read this genome as a source
 GENOMES=(
-  "genome-dev|Web development, TUI, Angular, software architecture|myorg/my-app"
+  "genome-dev|Web development, TUI, Angular, software architecture|myorg/my-app|no"
-  "genome-finance|Personal finance, investments, market analysis|"
+  "genome-finance|Personal finance, investments, market analysis||no"
-  "genome-homelab|Infrastructure, network configs, architecture logs|"
+  "genome-homelab|Infrastructure, network configs, architecture logs||no"
 )
 ```
 
@@ -692,7 +703,9 @@ If a key is lost or compromised:
 ```bash
 # From the knowledge-genome-orchestrator/ directory
 source lib/git-crypt.sh
-cd ~/knowledge-genome-orchestrator/genome-dev
+# If gcrypt_rotate_key operates on the CWD: cd into .../master-knowledge-genome/genome-dev
+# If it navigates by name instead:          cd into .../master-knowledge-genome
+cd ~/knowledge-genome-orchestrator/master-knowledge-genome
 gcrypt_rotate_key "genome-dev"
 ```
 
@@ -804,7 +817,10 @@ model must not waste context on:
 8. Appends the `INGEST | <slug>` entry to `wiki/log.md` (the model name comes from the
    orchestrator via `INGEST_MODEL` — the agent cannot reliably know its own tag)
 9. Runs scoped lint on exactly the pages touched this run (`scoped-lint.sh`, reusing
-   `lib/lint.sh`)
+   `lib/lint.sh`), including a **duplicate-slug advisory**: a slug created this run that is
+   highly similar to an entity/concept already in `wiki/index.md` is flagged in the PR so a
+   human can merge them. It is advisory only — it never fails the lint or blocks the PR
+   (threshold tunable via `KG_DUP_THRESHOLD`, default 70)
 10. Commits **only `wiki/`** on `feat/ai-ingest-<slug>` and opens a PR against the integration
     base (`INGEST_BASE`, default `main`); the body matches the `templates/pr-description.md`
     structure (Summary / Pages / Contradictions / Scoped Lint)
@@ -957,6 +973,10 @@ The agent proposes re-validation but does not change `maturity` without new sour
 
 ### Cross-genome references
 
+> **Status: planned.** The cross-genome collector and **navigation skill** described in this
+> section are specified but **not yet implemented** in this release — only the `ingest` skill
+> ships today. What follows documents the intended design and the boundary contract it will honour.
+
 Cross-domain knowledge moves by **pull, never push**: the genome you are working in draws
 material _in_; nothing is ever written into another genome. There are **no cross-genome
 wikilinks** — submodule pointers make relative paths brittle.
@@ -1059,7 +1079,7 @@ grep "^## \[" wiki/log.md | grep "CONFLICT"  # All conflicts
 grep "^## \[2026-05" wiki/log.md             # Entries from a specific month
 ```
 
-The orchestrator always injects only `tail -n 20 wiki/log.md` into agent context.
+ingest-semantic.py receives source text + existing entity/concept names (from index) as prompt context.
 The LLM never loads the full log.
 
 ---
@@ -1119,6 +1139,8 @@ Note: `.obsidian/` is in `.gitignore`. Workspace and plugin settings are local
 
 ### n8n automation
 
+n8n → SSH → ingest-semantic.py <genome> <raw> → run-ingest.sh <genome>.
+
 n8n (running on the storage node) can automate the ingest pipeline:
 
 1. Forgejo webhook fires on push to a genome's `raw/` directory

deploy/n8n/genome-PR-review.json

@@ -0,0 +1,773 @@
+{
+  "name": "Genome: PR review",
+  "nodes": [
+    {
+      "parameters": {
+        "httpMethod": "POST",
+        "path": "forgejo-pr-review-23319ab8687b16f10e0f278fb920c112",
+        "options": {}
+      },
+      "id": "58df1ca9-e48e-4834-b231-d97c974cd01b",
+      "name": "Webhook PR Review",
+      "type": "n8n-nodes-base.webhook",
+      "typeVersion": 2.1,
+      "position": [
+        2272,
+        1344
+      ],
+      "webhookId": "61ff3a5baa304571"
+    },
+    {
+      "parameters": {
+        "mode": "runOnceForEachItem",
+        "jsCode": "// THE only parser of the review side: parse the directive, VALIDATE, prepare the rework payload.\n// Security: only allow-listed maintainers may drive the gate; destructive directives require a\n// feat/ai-ingest-* branch on the expected base; raw_source is recovered from a machine-readable\n// marker that run-ingest.sh writes into the PR body.\nconst ALLOWED_SENDERS = ['Keru'];   // <-- maintainers allowed to issue directives\nconst BASE = 'develop';\n\n// n8n Run Once for Each Item: $json is the current webhook payload\nconst j = $json.body || $json;\nif (!j || typeof j !== 'object') {\n  return { directive: 'INVALID', reason: 'malformed webhook payload' };\n}\n\nconst review = j.review || null;\nconst comment = j.comment || null;\nconst pr = j.pull_request || j.issue || null;\n\n// Extract directive text from review content or comment body\nconst body = String(\n  (review && review.content) ||\n  (comment && comment.body) ||\n  ''\n);\nconst sender = String((j.sender && j.sender.login) || 'unknown');\n\n// Match directive at the start of the text (case-insensitive)\nconst m = body.match(/^\\s*(REWORK|RESTART|REVERT\\s+\\d+|SPLIT|REJECT|MERGE)\\s*:?/i);\nif (!m) return { directive: 'NONE' };\n\nconst headTok = m[1].toUpperCase().replace(/\\s+/g, ' ');\nconst directive = headTok.startsWith('REVERT') ? 'REVERT' : headTok;\nconst feedback = body.slice(m[0].length).trim() || '(nessun dettaglio fornito)';\n\n// Extract PR metadata safely\nconst prNumber = (pr && pr.number) || null;\nconst branch = (pr && pr.head && pr.head.ref) || null;\nconst base = (pr && pr.base && pr.base.ref) || null;\nconst repo = (pr && pr.base && pr.base.repo && pr.base.repo.name) ||\n             (j.repository && j.repository.name) || null;\nconst owner = (pr && pr.base && pr.base.repo && pr.base.repo.owner && pr.base.repo.owner.login) ||\n              (j.repository && j.repository.owner && j.repository.owner.login) || null;\nconst prBody = (pr && pr.body) || (j.issue && j.issue.body) || '';\n\n// Recover raw_source from machine-readable marker: <!-- kg:raw=path -->\n// Restricted to valid path characters, no spaces, no HTML breaking\nconst rawMatch = prBody.match(/<!--\\s*kg:raw=([^\\s>]+)\\s*-->/);\nconst raw = rawMatch ? rawMatch[1] : null;\n\n// REVERT is reserved for future Step 7 implementation\nif (directive === 'REVERT') {\n  return { directive: 'NONE', note: 'REVERT reserved for Step 7' };\n}\n\n// Authorization gate\nif (!ALLOWED_SENDERS.includes(sender)) {\n  return {\n    directive: 'UNAUTHORIZED',\n    attempted: directive,\n    sender,\n    prNumber,\n    owner,\n    repo\n  };\n}\n\n// Validation rules\nconst okGenome = !!repo && /^[a-z0-9][a-z0-9-]{0,63}$/.test(repo);\nconst okPr = !!prNumber && /^[0-9]+$/.test(String(prNumber));\nconst okBranch = !!branch && /^feat\\/ai-ingest-[a-z0-9-]+$/.test(branch);\nconst okBase = base === BASE;\nconst okRaw = (directive === 'MERGE')\n  ? true\n  : (!!raw && raw.startsWith('raw/') && !raw.includes('..') && /^[A-Za-z0-9._\\/-]+$/.test(raw));\n\nif (!okGenome || !okPr || !okBase || (directive !== 'MERGE' && !okBranch) || !okRaw) {\n  return {\n    directive: 'INVALID',\n    attempted: directive,\n    prNumber,\n    owner,\n    repo,\n    why: { okGenome, okPr, okBranch, okBase, okRaw }\n  };\n}\n\n// Encode feedback for safe transport through SSH/scripts\nconst feedback_b64 = Buffer.from(feedback, 'utf8').toString('base64');\n\nreturn {\n  directive,\n  prNumber,\n  branch,\n  base,\n  repo,\n  owner,\n  sender,\n  raw,\n  feedback,\n  feedback_b64\n};"
+      },
+      "id": "c668f595-0a28-4bd3-9125-22fee9350d78",
+      "name": "Parse & validate",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [
+        2496,
+        1344
+      ]
+    },
+    {
+      "parameters": {
+        "rules": {
+          "values": [
+            {
+              "conditions": {
+                "options": {
+                  "caseSensitive": true,
+                  "typeValidation": "strict",
+                  "version": 2
+                },
+                "conditions": [
+                  {
+                    "leftValue": "={{ $json.directive }}",
+                    "rightValue": "MERGE",
+                    "operator": {
+                      "type": "string",
+                      "operation": "equals"
+                    },
+                    "id": "4960f0868bc54687"
+                  }
+                ],
+                "combinator": "and"
+              }
+            },
+            {
+              "conditions": {
+                "options": {
+                  "caseSensitive": true,
+                  "typeValidation": "strict",
+                  "version": 2
+                },
+                "conditions": [
+                  {
+                    "leftValue": "={{ $json.directive }}",
+                    "rightValue": "REWORK",
+                    "operator": {
+                      "type": "string",
+                      "operation": "equals"
+                    },
+                    "id": "34002fdd92834d38"
+                  }
+                ],
+                "combinator": "and"
+              }
+            },
+            {
+              "conditions": {
+                "options": {
+                  "caseSensitive": true,
+                  "typeValidation": "strict",
+                  "version": 2
+                },
+                "conditions": [
+                  {
+                    "leftValue": "={{ $json.directive }}",
+                    "rightValue": "RESTART",
+                    "operator": {
+                      "type": "string",
+                      "operation": "equals"
+                    },
+                    "id": "d412a74e32ac4f0c"
+                  }
+                ],
+                "combinator": "and"
+              }
+            },
+            {
+              "conditions": {
+                "options": {
+                  "caseSensitive": true,
+                  "typeValidation": "strict",
+                  "version": 2
+                },
+                "conditions": [
+                  {
+                    "leftValue": "={{ $json.directive }}",
+                    "rightValue": "SPLIT",
+                    "operator": {
+                      "type": "string",
+                      "operation": "equals"
+                    },
+                    "id": "c0810b33fa474ca0"
+                  }
+                ],
+                "combinator": "and"
+              }
+            },
+            {
+              "conditions": {
+                "options": {
+                  "caseSensitive": true,
+                  "typeValidation": "strict",
+                  "version": 2
+                },
+                "conditions": [
+                  {
+                    "leftValue": "={{ $json.directive }}",
+                    "rightValue": "REJECT",
+                    "operator": {
+                      "type": "string",
+                      "operation": "equals"
+                    },
+                    "id": "531039e699c44cea"
+                  }
+                ],
+                "combinator": "and"
+              }
+            },
+            {
+              "conditions": {
+                "options": {
+                  "caseSensitive": true,
+                  "typeValidation": "strict",
+                  "version": 2
+                },
+                "conditions": [
+                  {
+                    "leftValue": "={{ $json.directive }}",
+                    "rightValue": "UNAUTHORIZED",
+                    "operator": {
+                      "type": "string",
+                      "operation": "equals"
+                    },
+                    "id": "cfbd691d2e9a4c2a"
+                  }
+                ],
+                "combinator": "and"
+              }
+            },
+            {
+              "conditions": {
+                "options": {
+                  "caseSensitive": true,
+                  "typeValidation": "strict",
+                  "version": 2
+                },
+                "conditions": [
+                  {
+                    "leftValue": "={{ $json.directive }}",
+                    "rightValue": "INVALID",
+                    "operator": {
+                      "type": "string",
+                      "operation": "equals"
+                    },
+                    "id": "251f5b7beea6424a"
+                  }
+                ],
+                "combinator": "and"
+              }
+            }
+          ]
+        },
+        "options": {
+          "fallbackOutput": "none"
+        }
+      },
+      "id": "489736cc-bab6-4664-8087-91b6d9ff31ad",
+      "name": "Switch",
+      "type": "n8n-nodes-base.switch",
+      "typeVersion": 3.4,
+      "position": [
+        2736,
+        1344
+      ]
+    },
+    {
+      "parameters": {
+        "method": "POST",
+        "url": "=https://git.keruhomelab.com/api/v1/repos/{{ $('Parse & validate').first().json.owner }}/{{ $('Parse & validate').first().json.repo }}/pulls/{{ $('Parse & validate').first().json.prNumber }}/merge",
+        "authentication": "genericCredentialType",
+        "genericAuthType": "httpHeaderAuth",
+        "sendBody": true,
+        "specifyBody": "json",
+        "jsonBody": "={\n  \"Do\": \"merge\"\n}",
+        "options": {
+          "timeout": 15000
+        }
+      },
+      "id": "3440cb8d-ae4c-4523-ae13-ee5667d24252",
+      "name": "Forgejo Merge PR",
+      "type": "n8n-nodes-base.httpRequest",
+      "typeVersion": 4.4,
+      "position": [
+        2976,
+        1104
+      ],
+      "credentials": {
+        "httpHeaderAuth": {
+          "id": "TBPXSWOF63k9mvm8",
+          "name": "ntfy-token"
+        }
+      }
+    },
+    {
+      "parameters": {
+        "conditions": {
+          "options": {
+            "caseSensitive": true,
+            "typeValidation": "loose",
+            "version": 2
+          },
+          "conditions": [
+            {
+              "id": "cc369b5fc3d246a4",
+              "leftValue": "={{ $('Parse & validate').first().json.branch }}",
+              "rightValue": "feat/ai-ingest-",
+              "operator": {
+                "type": "string",
+                "operation": "startsWith"
+              }
+            }
+          ],
+          "combinator": "and"
+        },
+        "options": {}
+      },
+      "id": "e6d45fce-83d0-44ca-9fa4-86558fec1a0f",
+      "name": "Guardia feat/",
+      "type": "n8n-nodes-base.if",
+      "typeVersion": 2.2,
+      "position": [
+        2976,
+        1328
+      ]
+    },
+    {
+      "parameters": {
+        "method": "PATCH",
+        "url": "=https://git.keruhomelab.com/api/v1/repos/{{ $('Parse & validate').first().json.owner }}/{{ $('Parse & validate').first().json.repo }}/pulls/{{ $('Parse & validate').first().json.prNumber }}",
+        "authentication": "genericCredentialType",
+        "genericAuthType": "httpHeaderAuth",
+        "sendBody": true,
+        "specifyBody": "json",
+        "jsonBody": "={\n  \"state\": \"closed\"\n}",
+        "options": {
+          "timeout": 15000
+        }
+      },
+      "id": "1601f705-c758-4df6-a3bd-e3ac2e202c94",
+      "name": "Forgejo Close PR",
+      "type": "n8n-nodes-base.httpRequest",
+      "typeVersion": 4.4,
+      "position": [
+        3200,
+        1296
+      ],
+      "credentials": {
+        "httpHeaderAuth": {
+          "id": "TBPXSWOF63k9mvm8",
+          "name": "ntfy-token"
+        }
+      }
+    },
+    {
+      "parameters": {
+        "method": "DELETE",
+        "url": "=https://git.keruhomelab.com/api/v1/repos/{{ $('Parse & validate').first().json.owner }}/{{ $('Parse & validate').first().json.repo }}/branches/{{ encodeURIComponent($('Parse & validate').first().json.branch) }}",
+        "authentication": "genericCredentialType",
+        "genericAuthType": "httpHeaderAuth",
+        "options": {
+          "timeout": 15000
+        }
+      },
+      "id": "c2ff2247-efe1-4809-a435-9973188d61bb",
+      "name": "Forgejo Delete Branch",
+      "type": "n8n-nodes-base.httpRequest",
+      "typeVersion": 4.4,
+      "position": [
+        3424,
+        1296
+      ],
+      "credentials": {
+        "httpHeaderAuth": {
+          "id": "TBPXSWOF63k9mvm8",
+          "name": "ntfy-token"
+        }
+      }
+    },
+    {
+      "parameters": {
+        "conditions": {
+          "options": {
+            "caseSensitive": true,
+            "leftValue": "",
+            "typeValidation": "strict",
+            "version": 2
+          },
+          "conditions": [
+            {
+              "id": "55cf6c2a6c7d4d79",
+              "leftValue": "={{ $('Parse & validate').first().json.directive }}",
+              "rightValue": "REJECT",
+              "operator": {
+                "type": "string",
+                "operation": "equals"
+              }
+            }
+          ],
+          "combinator": "and"
+        },
+        "options": {}
+      },
+      "id": "a1dbbc06-555d-4a1d-8fbf-ee75f617e98a",
+      "name": "E' REJECT?",
+      "type": "n8n-nodes-base.if",
+      "typeVersion": 2.2,
+      "position": [
+        3648,
+        1296
+      ]
+    },
+    {
+      "parameters": {
+        "workflowId": {
+          "__rl": true,
+          "value": "zbtRXWsLt56nEIfz",
+          "mode": "list",
+          "cachedResultUrl": "/workflow/zbtRXWsLt56nEIfz",
+          "cachedResultName": "Power Manager"
+        },
+        "workflowInputs": {
+          "mappingMode": "defineBelow",
+          "value": {
+            "mode": "ensure-on"
+          },
+          "matchingColumns": [
+            "mode"
+          ],
+          "schema": [
+            {
+              "id": "mode",
+              "displayName": "mode",
+              "required": false,
+              "defaultMatch": false,
+              "display": true,
+              "canBeUsedToMatch": true,
+              "type": "string",
+              "removed": false
+            }
+          ],
+          "attemptToConvertTypes": false,
+          "convertFieldsToString": true
+        },
+        "options": {}
+      },
+      "id": "7fc3e648-4712-4eef-a6f3-12c8805ade1f",
+      "name": "Power Manager - ensure-on",
+      "type": "n8n-nodes-base.executeWorkflow",
+      "typeVersion": 1.3,
+      "position": [
+        3648,
+        1168
+      ]
+    },
+    {
+      "parameters": {
+        "workflowId": {
+          "__rl": true,
+          "value": "VIi2ovb5gJxNJLbg",
+          "mode": "list",
+          "cachedResultUrl": "/workflow/VIi2ovb5gJxNJLbg",
+          "cachedResultName": "Genome: run-one-ingest"
+        },
+        "workflowInputs": {
+          "mappingMode": "defineBelow",
+          "value": {
+            "genome": "={{ $('Parse & validate').first().json.repo }}",
+            "raw": "={{ $('Parse & validate').first().json.raw }}",
+            "mode": "rework",
+            "feedback_b64": "={{ $('Parse & validate').first().json.feedback_b64 }}",
+            "reason": "={{ $('Parse & validate').first().json.directive }}",
+            "prevPr": "={{ String($('Parse & validate').first().json.prNumber || '') }}"
+          },
+          "matchingColumns": [],
+          "schema": [
+            {
+              "id": "genome",
+              "displayName": "genome",
+              "required": false,
+              "defaultMatch": false,
+              "display": true,
+              "canBeUsedToMatch": true,
+              "type": "string",
+              "removed": false
+            },
+            {
+              "id": "raw",
+              "displayName": "raw",
+              "required": false,
+              "defaultMatch": false,
+              "display": true,
+              "canBeUsedToMatch": true,
+              "type": "string",
+              "removed": false
+            },
+            {
+              "id": "mode",
+              "displayName": "mode",
+              "required": false,
+              "defaultMatch": false,
+              "display": true,
+              "canBeUsedToMatch": true,
+              "type": "string",
+              "removed": false
+            },
+            {
+              "id": "feedback_b64",
+              "displayName": "feedback_b64",
+              "required": false,
+              "defaultMatch": false,
+              "display": true,
+              "canBeUsedToMatch": true,
+              "type": "string",
+              "removed": false
+            },
+            {
+              "id": "reason",
+              "displayName": "reason",
+              "required": false,
+              "defaultMatch": false,
+              "display": true,
+              "canBeUsedToMatch": true,
+              "type": "string",
+              "removed": false
+            },
+            {
+              "id": "prevPr",
+              "displayName": "prevPr",
+              "required": false,
+              "defaultMatch": false,
+              "display": true,
+              "canBeUsedToMatch": true,
+              "type": "string",
+              "removed": false
+            }
+          ],
+          "attemptToConvertTypes": false,
+          "convertFieldsToString": true
+        },
+        "options": {
+          "waitForSubWorkflow": false
+        }
+      },
+      "id": "9704c050-5c63-49fd-a26d-efbae9d92175",
+      "name": "Run one ingest (rework)",
+      "type": "n8n-nodes-base.executeWorkflow",
+      "typeVersion": 1.3,
+      "position": [
+        3856,
+        1168
+      ]
+    },
+    {
+      "parameters": {
+        "mode": "runOnceForEachItem",
+        "jsCode": "// merged (MERGE) / closed (REJECT). The HTTP node replaced $json with the API response,\n// so we read context from the parser via node reference (single review -> .first() is safe).\n// Fallback values prevent crashes if the parser node is unreachable.\nconst p = $('Parse & validate').first().json || {};\nconst repo = p.repo || 'unknown';\nconst owner = p.owner || 'unknown';\nconst prNumber = p.prNumber || '?';\nconst base = p.base || 'develop';\nconst branch = p.branch || 'unknown';\nconst sender = p.sender || 'unknown';\nconst directive = p.directive || 'UNKNOWN';\nconst feedback = p.feedback || '';\n\nconst repoUrl = (owner && repo && repo !== 'unknown')\n  ? `https://git.keruhomelab.com/${owner}/${repo}`\n  : '';\nconst prUrl = (repoUrl && prNumber !== '?')\n  ? `${repoUrl}/pulls/${prNumber}`\n  : '';\n\nlet n;\nif (directive === 'MERGE') {\n  n = {\n    topic: 'genome-ingest',\n    title: `${repo} · PR #${prNumber} mergiata`,\n    priority: 'default',\n    tags: 'twisted_rightwards_arrows',\n    click: prUrl,\n    actions: `view, Vedi la PR, ${prUrl}`,\n    body: `PR #${prNumber} mergiata su \\`${base}\\` da **${sender}**.`\n  };\n} else {\n  n = {\n    topic: 'genome-ingest',\n    title: `${repo} · PR #${prNumber} chiusa`,\n    priority: 'default',\n    tags: 'wastebasket',\n    click: repoUrl,\n    actions: '',\n    body: `**REJECT** di **${sender}**: PR #${prNumber} chiusa e branch \\`${branch}\\` rimosso. Nessun nuovo tentativo.\\n> ${feedback}`\n  };\n}\n\nreturn n;"
+      },
+      "id": "1ce634fd-d402-4a84-9ba1-04673ddffce9",
+      "name": "Build ntfy action",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [
+        3856,
+        1344
+      ]
+    },
+    {
+      "parameters": {
+        "mode": "runOnceForEachItem",
+        "jsCode": "// Security / near-miss: unauthorized sender, invalid directive, or the feat/ guard.\n// On all three paths Switch/Guardia pass the parser output through, so $json carries the directive + context.\nconst d = $json || {};\nconst directive = d.directive || 'UNKNOWN';\nconst attempted = d.attempted || directive;\nconst sender = d.sender || 'unknown';\nconst prNumber = d.prNumber || '?';\nconst branch = d.branch || 'unknown';\nconst owner = d.owner || '';\nconst repo = d.repo || '';\n\nconst repoUrl = (owner && repo) ? `https://git.keruhomelab.com/${owner}/${repo}` : '';\n\nlet n;\nif (directive === 'UNAUTHORIZED') {\n  n = {\n    topic: 'genome-ingest',\n    title: `Sicurezza · direttiva non autorizzata`,\n    priority: 'high',\n    tags: 'no_entry',\n    click: repoUrl,\n    actions: '',\n    body: `**${sender}** ha tentato \\`${attempted}\\` su PR #${prNumber}, ma non è tra i maintainer autorizzati. **Nessuna azione** eseguita.`\n  };\n} else if (directive === 'INVALID') {\n  n = {\n    topic: 'genome-ingest',\n    title: `Direttiva non applicata`,\n    priority: 'low',\n    tags: 'information_source',\n    click: repoUrl,\n    actions: '',\n    body: `\\`${attempted}\\` su PR #${prNumber} ignorata: precondizioni non soddisfatte (branch / base / marker raw).`\n  };\n} else {\n  // Guardia feat/ false branch: destructive action on a non-feat/ai-ingest-* branch\n  n = {\n    topic: 'genome-ingest',\n    title: `Sicurezza · branch protetto`,\n    priority: 'high',\n    tags: 'no_entry',\n    click: repoUrl,\n    actions: '',\n    body: `Rifiutata azione distruttiva (\\`${attempted || directive}\\`) sul branch \\`${branch}\\`: non è un \\`feat/ai-ingest-*\\`. **Nessuna modifica.**`\n  };\n}\n\nreturn n;"
+      },
+      "id": "32b16592-5126-4cc2-a3f2-d1bda58ac724",
+      "name": "Build ntfy sicurezza",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [
+        3200,
+        1536
+      ]
+    },
+    {
+      "parameters": {
+        "method": "POST",
+        "url": "=http://ntfy/{{ $json.topic }}",
+        "authentication": "genericCredentialType",
+        "genericAuthType": "httpBearerAuth",
+        "sendHeaders": true,
+        "headerParameters": {
+          "parameters": [
+            {
+              "name": "Title",
+              "value": "={{ $json.title }}"
+            },
+            {
+              "name": "Priority",
+              "value": "={{ $json.priority }}"
+            },
+            {
+              "name": "Tags",
+              "value": "={{ $json.tags }}"
+            },
+            {
+              "name": "Click",
+              "value": "={{ $json.click }}"
+            },
+            {
+              "name": "Actions",
+              "value": "={{ $json.actions }}"
+            },
+            {
+              "name": "Markdown",
+              "value": "yes"
+            }
+          ]
+        },
+        "sendBody": true,
+        "contentType": "raw",
+        "rawContentType": "Raw / Text",
+        "body": "={{ $json.body }}",
+        "options": {
+          "timeout": 15000
+        }
+      },
+      "id": "4d45b486-de42-4c7f-be21-b5bfbc05fd44",
+      "name": "ntfy: send",
+      "type": "n8n-nodes-base.httpRequest",
+      "typeVersion": 4.4,
+      "position": [
+        4080,
+        1424
+      ],
+      "credentials": {
+        "httpHeaderAuth": {
+          "id": "TBPXSWOF63k9mvm8",
+          "name": "ntfy-token"
+        },
+        "httpBearerAuth": {
+          "id": "nCv4CUN7Ef086Ewj",
+          "name": "Bearer Auth account"
+        }
+      }
+    }
+  ],
+  "pinData": {},
+  "connections": {
+    "Webhook PR Review": {
+      "main": [
+        [
+          {
+            "node": "Parse & validate",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Parse & validate": {
+      "main": [
+        [
+          {
+            "node": "Switch",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Switch": {
+      "main": [
+        [
+          {
+            "node": "Forgejo Merge PR",
+            "type": "main",
+            "index": 0
+          }
+        ],
+        [
+          {
+            "node": "Power Manager - ensure-on",
+            "type": "main",
+            "index": 0
+          }
+        ],
+        [
+          {
+            "node": "Guardia feat/",
+            "type": "main",
+            "index": 0
+          }
+        ],
+        [
+          {
+            "node": "Guardia feat/",
+            "type": "main",
+            "index": 0
+          }
+        ],
+        [
+          {
+            "node": "Guardia feat/",
+            "type": "main",
+            "index": 0
+          }
+        ],
+        [
+          {
+            "node": "Build ntfy sicurezza",
+            "type": "main",
+            "index": 0
+          }
+        ],
+        [
+          {
+            "node": "Build ntfy sicurezza",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Forgejo Merge PR": {
+      "main": [
+        [
+          {
+            "node": "Build ntfy action",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Guardia feat/": {
+      "main": [
+        [
+          {
+            "node": "Forgejo Close PR",
+            "type": "main",
+            "index": 0
+          }
+        ],
+        [
+          {
+            "node": "Build ntfy sicurezza",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Forgejo Close PR": {
+      "main": [
+        [
+          {
+            "node": "Forgejo Delete Branch",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Forgejo Delete Branch": {
+      "main": [
+        [
+          {
+            "node": "E' REJECT?",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "E' REJECT?": {
+      "main": [
+        [
+          {
+            "node": "Build ntfy action",
+            "type": "main",
+            "index": 0
+          }
+        ],
+        [
+          {
+            "node": "Power Manager - ensure-on",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Power Manager - ensure-on": {
+      "main": [
+        [
+          {
+            "node": "Run one ingest (rework)",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Build ntfy action": {
+      "main": [
+        [
+          {
+            "node": "ntfy: send",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Build ntfy sicurezza": {
+      "main": [
+        [
+          {
+            "node": "ntfy: send",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    }
+  },
+  "active": true,
+  "settings": {
+    "executionOrder": "v1",
+    "binaryMode": "separate",
+    "timeSavedMode": "fixed",
+    "errorWorkflow": "7Vws3gCX3QnjM3oD",
+    "callerPolicy": "workflowsFromSameOwner",
+    "availableInMCP": false
+  },
+  "versionId": "22998a54-cd9a-4b57-9c80-df97085a997c",
+  "meta": {
+    "instanceId": "96b2f0ec76a4400bbd481c617b24b3b87024cc7a913efacccaf9fc85722e7417"
+  },
+  "id": "iho7kFQsXbGIxG7P",
+  "tags": []
+}

deploy/n8n/genome-ingest-manuale.json

@@ -0,0 +1,170 @@
+{
+  "name": "Genome: ingest MANUALE (scratch)",
+  "nodes": [
+    {
+      "parameters": {},
+      "type": "n8n-nodes-base.manualTrigger",
+      "typeVersion": 1,
+      "position": [
+        0,
+        0
+      ],
+      "id": "2101e704-6275-419d-9963-29a142e5811c",
+      "name": "Esegui manualmente"
+    },
+    {
+      "parameters": {
+        "authentication": "privateKey",
+        "command": "ssh vm101 'pi ingest genome-test raw/articles/il-grano-saraceno.md'"
+      },
+      "type": "n8n-nodes-base.ssh",
+      "typeVersion": 1,
+      "position": [
+        224,
+        0
+      ],
+      "id": "8ade2def-2d53-4860-88a5-2ca734c6e54a",
+      "name": "SSH: pi ingest (manuale)",
+      "credentials": {
+        "sshPrivateKey": {
+          "id": "GJQjKzte7Hjdfz89",
+          "name": "n8n container -> n8n-runner@nexus"
+        }
+      }
+    },
+    {
+      "parameters": {
+        "mode": "runOnceForEachItem",
+        "jsCode": "// ultima riga JSON di run-ingest.sh (ha 'run_id=' davanti)\nconst out = ($json.stdout || '').trim();\nconst line = out.split('\\n').filter(l => l.trim().startsWith('{')).pop();\nif (!line) return { status: 'error', reason: 'nessuna riga JSON run-ingest', raw: out };\ntry { return JSON.parse(line); } catch (e) { return { status: 'error', reason: 'JSON non parsabile', raw: line }; }"
+      },
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [
+        448,
+        0
+      ],
+      "id": "d84cdeaf-612a-454c-8b4d-31824ae6d71e",
+      "name": "Parse ingest"
+    },
+    {
+      "parameters": {
+        "mode": "runOnceForEachItem",
+        "jsCode": "const d=$json;let n;\nif (d.status==='ok'){\n  n={title:`Ingest ${d.slug}: PR aperta`,priority:'default',tags:'inbox_tray',\n     body:`\\u2705 ${d.slug}: PR aperta (lint ${d.lint_clean?'clean':'KO'}${d.conflict?', CONFLITTO':''})\\n\\n\\ud83d\\udd17 ${d.pr_url}`};\n} else if (d.status==='pr_failed'){\n  n={title:`Ingest ${d.slug}: PR FALLITA`,priority:'high',tags:'warning',\n     body:`\\u26a0\\ufe0f ${d.slug}: semantic/lint ok ma PR non aperta.\\n\\n${(d.detail||'').split('\\n')[0]}`};\n} else {\n  n={title:'Ingest: ERRORE',priority:'high',tags:'rotating_light',\n     body:`\\u274c ${d.reason||'errore'}\\n\\n${(d.raw||'').slice(0,300)}`};\n}\nreturn n;"
+      },
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [
+        672,
+        0
+      ],
+      "id": "eadd9275-b38c-416b-b15e-0999f70a05fb",
+      "name": "Build ntfy"
+    },
+    {
+      "parameters": {
+        "method": "POST",
+        "url": "http://ntfy/homelab-genome",
+        "authentication": "genericCredentialType",
+        "genericAuthType": "httpBearerAuth",
+        "sendHeaders": true,
+        "headerParameters": {
+          "parameters": [
+            {
+              "name": "Title",
+              "value": "={{ $json.title }}"
+            },
+            {
+              "name": "Priority",
+              "value": "={{ $json.priority }}"
+            },
+            {
+              "name": "Tags",
+              "value": "={{ $json.tags }}"
+            }
+          ]
+        },
+        "sendBody": true,
+        "contentType": "raw",
+        "rawContentType": "Raw / Text",
+        "body": "={{ $json.body }}",
+        "options": {}
+      },
+      "type": "n8n-nodes-base.httpRequest",
+      "typeVersion": 4.4,
+      "position": [
+        880,
+        0
+      ],
+      "id": "63ab577b-893a-4b3d-8f13-b377be778099",
+      "name": "ntfy: send notification",
+      "credentials": {
+        "httpHeaderAuth": {
+          "id": "TBPXSWOF63k9mvm8",
+          "name": "ntfy-token"
+        },
+        "httpBearerAuth": {
+          "id": "nCv4CUN7Ef086Ewj",
+          "name": "Bearer Auth account"
+        }
+      }
+    }
+  ],
+  "pinData": {},
+  "connections": {
+    "Esegui manualmente": {
+      "main": [
+        [
+          {
+            "node": "SSH: pi ingest (manuale)",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "SSH: pi ingest (manuale)": {
+      "main": [
+        [
+          {
+            "node": "Parse ingest",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Parse ingest": {
+      "main": [
+        [
+          {
+            "node": "Build ntfy",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Build ntfy": {
+      "main": [
+        [
+          {
+            "node": "ntfy: send notification",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    }
+  },
+  "active": false,
+  "settings": {
+    "executionOrder": "v1",
+    "binaryMode": "separate"
+  },
+  "versionId": "df06ce3b-1ea8-43be-91ff-02c77972cfe2",
+  "meta": {
+    "instanceId": "96b2f0ec76a4400bbd481c617b24b3b87024cc7a913efacccaf9fc85722e7417"
+  },
+  "id": "RNoSaRLYG9vcMn6M",
+  "tags": []
+}

deploy/n8n/genome-ingest.json

@@ -0,0 +1,419 @@
+{
+  "name": "Genome: ingest",
+  "nodes": [
+    {
+      "parameters": {
+        "httpMethod": "POST",
+        "path": "forgejo-push",
+        "options": {}
+      },
+      "id": "8c44b478-1a95-4c3b-8ac1-d7c57e228414",
+      "name": "Webhook",
+      "type": "n8n-nodes-base.webhook",
+      "typeVersion": 2.1,
+      "position": [
+        1520,
+        1728
+      ],
+      "webhookId": "cf215f5d31e04dd2"
+    },
+    {
+      "parameters": {
+        "jsCode": "// Bell filter: proceed ONLY on develop pushes that actually touch raw/.\n// Returning [] stops the flow (no node needed).\n// Performance: never wake vm101 for wiki-only pushes (e.g. an ingest PR merged back to develop).\n// pending-raw remains the source of truth.\nconst item = $input.first().json;\nconst b = item.body || item;\nconst ref = String(b.ref || '');\nconst genome = String((b.repository && b.repository.name) || '').toLowerCase().trim();\n\n// Branch filter\nif (ref !== 'refs/heads/develop') return [];\n\n// Genome name validation (DNS-like: lowercase alphanum + hyphen, 1-64 chars)\nif (!/^[a-z0-9][a-z0-9-]{0,63}$/.test(genome)) return [];\n\n// Collect all touched paths safely (added, modified, removed)\nconst commits = Array.isArray(b.commits) ? b.commits : [];\nconst touched = [];\nfor (const c of commits) {\n  if (!c || typeof c !== 'object') continue;\n  for (const key of ['added', 'modified', 'removed']) {\n    const list = c[key];\n    if (!Array.isArray(list)) continue;\n    for (const p of list) {\n      if (typeof p === 'string' && p.startsWith('raw/')) {\n        touched.push(p);\n      }\n    }\n  }\n}\n\n// Gate: stop if nothing under raw/ was touched\nif (touched.length === 0) return [];\n\nreturn [{ json: { genome, touchedCount: touched.length } }];"
+      },
+      "id": "604787c7-4e83-468e-9a98-3ac084203040",
+      "name": "Gate push",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [
+        1744,
+        1728
+      ]
+    },
+    {
+      "parameters": {
+        "workflowId": {
+          "__rl": true,
+          "value": "zbtRXWsLt56nEIfz",
+          "mode": "list",
+          "cachedResultUrl": "/workflow/zbtRXWsLt56nEIfz",
+          "cachedResultName": "Power Manager"
+        },
+        "workflowInputs": {
+          "mappingMode": "defineBelow",
+          "value": {
+            "mode": "ensure-on"
+          },
+          "matchingColumns": [
+            "mode"
+          ],
+          "schema": [
+            {
+              "id": "mode",
+              "displayName": "mode",
+              "required": false,
+              "defaultMatch": false,
+              "display": true,
+              "canBeUsedToMatch": true,
+              "type": "string",
+              "removed": false
+            }
+          ],
+          "attemptToConvertTypes": false,
+          "convertFieldsToString": true
+        },
+        "options": {}
+      },
+      "id": "f93073a3-7753-4ce1-9ef1-2a0c16386543",
+      "name": "Power Manager - ensure-on",
+      "type": "n8n-nodes-base.executeWorkflow",
+      "typeVersion": 1.3,
+      "position": [
+        1952,
+        1728
+      ]
+    },
+    {
+      "parameters": {
+        "authentication": "privateKey",
+        "command": "=ssh vm101 'pi pending-raw {{ $('Gate push').first().json.genome }}'"
+      },
+      "id": "876dbdaf-3620-4c2c-a65b-336f0b11198c",
+      "name": "SSH: pending-raw",
+      "type": "n8n-nodes-base.ssh",
+      "typeVersion": 1,
+      "position": [
+        2176,
+        1728
+      ],
+      "credentials": {
+        "sshPrivateKey": {
+          "id": "GJQjKzte7Hjdfz89",
+          "name": "n8n container -> n8n-runner@nexus"
+        }
+      }
+    },
+    {
+      "parameters": {
+        "jsCode": "// Parse pending-raw -> one item per raw, carrying everything run-one-ingest needs.\n// Unsafe filenames (spaces / odd chars) are NOT ingested -> a 'badname' item -> ntfy.\nconst out = String($input.first().json.stdout || '').trim();\nlet d;\ntry {\n  d = JSON.parse(out);\n} catch (e) {\n  return [{ json: { _kind: 'error', reason: 'pending-raw non parsabile', raw: out.substring(0, 500) } }];\n}\n\nif (!d || typeof d !== 'object') {\n  return [{ json: { _kind: 'error', reason: 'pending-raw non è un oggetto JSON', raw: out.substring(0, 500) } }];\n}\n\nconst files = Array.isArray(d.files) ? d.files : [];\nif (files.length === 0) return [];\n\n// Build reason map from detail array\nconst why = {};\nfor (const it of (Array.isArray(d.detail) ? d.detail : [])) {\n  if (it && typeof it.path === 'string' && typeof it.reason === 'string') {\n    why[it.path] = it.reason;\n  }\n}\n\nconst SAFE = /^[A-Za-z0-9._\\/-]+$/;\nconst items = [];\nfor (const raw of files) {\n  if (typeof raw !== 'string') {\n    items.push({ json: { _kind: 'badname', genome: d.genome, raw: String(raw),\n      hint: String(raw).replace(/[^A-Za-z0-9._\\/-]+/g, '-').toLowerCase() || 'invalid' } });\n    continue;\n  }\n  if (SAFE.test(raw)) {\n    items.push({ json: { _kind: 'ingest', genome: d.genome, raw,\n      mode: 'ingest', feedback_b64: '', reason: why[raw] || 'new', prevPr: '' } });\n  } else {\n    const hint = raw.replace(/[^A-Za-z0-9._\\/-]+/g, '-').toLowerCase() || 'invalid';\n    items.push({ json: { _kind: 'badname', genome: d.genome, raw, hint } });\n  }\n}\nreturn items;"
+      },
+      "id": "f5bbbed3-222e-4129-a764-7cf47d69c5ce",
+      "name": "Split raw files",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [
+        2400,
+        1728
+      ]
+    },
+    {
+      "parameters": {
+        "conditions": {
+          "options": {
+            "caseSensitive": true,
+            "leftValue": "",
+            "typeValidation": "strict",
+            "version": 2
+          },
+          "conditions": [
+            {
+              "id": "cbacf5d98d594ba5",
+              "leftValue": "={{ $json._kind }}",
+              "rightValue": "ingest",
+              "operator": {
+                "type": "string",
+                "operation": "equals"
+              }
+            }
+          ],
+          "combinator": "and"
+        },
+        "options": {}
+      },
+      "id": "5398e2c4-c7ca-4ca4-a2d7-e75077453b7c",
+      "name": "Nome valido?",
+      "type": "n8n-nodes-base.if",
+      "typeVersion": 2.2,
+      "position": [
+        2624,
+        1728
+      ]
+    },
+    {
+      "parameters": {
+        "workflowId": {
+          "__rl": true,
+          "value": "VIi2ovb5gJxNJLbg",
+          "mode": "list",
+          "cachedResultUrl": "/workflow/VIi2ovb5gJxNJLbg",
+          "cachedResultName": "Genome: run-one-ingest"
+        },
+        "workflowInputs": {
+          "mappingMode": "defineBelow",
+          "value": {
+            "genome": "={{ $json.genome }}",
+            "raw": "={{ $json.raw }}",
+            "mode": "ingest",
+            "feedback_b64": "",
+            "reason": "={{ $json.reason }}",
+            "prevPr": ""
+          },
+          "matchingColumns": [],
+          "schema": [
+            {
+              "id": "genome",
+              "displayName": "genome",
+              "required": false,
+              "defaultMatch": false,
+              "display": true,
+              "canBeUsedToMatch": true,
+              "type": "string",
+              "removed": false
+            },
+            {
+              "id": "raw",
+              "displayName": "raw",
+              "required": false,
+              "defaultMatch": false,
+              "display": true,
+              "canBeUsedToMatch": true,
+              "type": "string",
+              "removed": false
+            },
+            {
+              "id": "mode",
+              "displayName": "mode",
+              "required": false,
+              "defaultMatch": false,
+              "display": true,
+              "canBeUsedToMatch": true,
+              "type": "string",
+              "removed": false
+            },
+            {
+              "id": "feedback_b64",
+              "displayName": "feedback_b64",
+              "required": false,
+              "defaultMatch": false,
+              "display": true,
+              "canBeUsedToMatch": true,
+              "type": "string",
+              "removed": false
+            },
+            {
+              "id": "reason",
+              "displayName": "reason",
+              "required": false,
+              "defaultMatch": false,
+              "display": true,
+              "canBeUsedToMatch": true,
+              "type": "string",
+              "removed": false
+            },
+            {
+              "id": "prevPr",
+              "displayName": "prevPr",
+              "required": false,
+              "defaultMatch": false,
+              "display": true,
+              "canBeUsedToMatch": true,
+              "type": "string",
+              "removed": false
+            }
+          ],
+          "attemptToConvertTypes": false,
+          "convertFieldsToString": true
+        },
+        "options": {
+          "waitForSubWorkflow": false
+        }
+      },
+      "id": "0f274662-62bb-448b-ae4b-47e4bbcfd35a",
+      "name": "Run one ingest",
+      "type": "n8n-nodes-base.executeWorkflow",
+      "typeVersion": 1.3,
+      "position": [
+        2832,
+        1616
+      ]
+    },
+    {
+      "parameters": {
+        "mode": "runOnceForEachItem",
+        "jsCode": "// Build ntfy notification for files with invalid names.\n// Run Once for Each Item: $json is the current badname item.\nconst d = $json || {};\nconst genome = d.genome || 'unknown';\nconst raw = String(d.raw || 'unknown');\nconst hint = String(d.hint || 'unknown');\n\n// Escape backticks to avoid breaking markdown\nconst rawEsc = raw.replace(/`/g, '\\`');\nconst hintEsc = hint.replace(/`/g, '\\`');\n\nreturn {\n  topic: 'genome-ingest',\n  title: `${genome} · file da rinominare`,\n  priority: 'high',\n  tags: 'warning',\n  click: '',\n  actions: '',\n  body: `Il file \\`${rawEsc}\\` ha spazi o caratteri non ammessi e **non** è stato ingerito.\\nRinominalo in: \\`${hintEsc}\\``\n};"
+      },
+      "id": "0f785bcd-cdc6-4dac-9ced-1c5cfa3453dc",
+      "name": "Build ntfy badname",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [
+        2832,
+        1840
+      ]
+    },
+    {
+      "parameters": {
+        "method": "POST",
+        "url": "=http://ntfy/{{ $json.topic }}",
+        "authentication": "genericCredentialType",
+        "genericAuthType": "httpBearerAuth",
+        "sendHeaders": true,
+        "headerParameters": {
+          "parameters": [
+            {
+              "name": "Title",
+              "value": "={{ $json.title }}"
+            },
+            {
+              "name": "Priority",
+              "value": "={{ $json.priority }}"
+            },
+            {
+              "name": "Tags",
+              "value": "={{ $json.tags }}"
+            },
+            {
+              "name": "Click",
+              "value": "={{ $json.click }}"
+            },
+            {
+              "name": "Actions",
+              "value": "={{ $json.actions }}"
+            },
+            {
+              "name": "Markdown",
+              "value": "yes"
+            }
+          ]
+        },
+        "sendBody": true,
+        "contentType": "raw",
+        "rawContentType": "Raw / Text",
+        "body": "={{ $json.body }}",
+        "options": {
+          "timeout": 15000
+        }
+      },
+      "id": "9cd2bde3-6846-4855-ad01-e3a4cdbce208",
+      "name": "ntfy: send",
+      "type": "n8n-nodes-base.httpRequest",
+      "typeVersion": 4.4,
+      "position": [
+        3056,
+        1840
+      ],
+      "credentials": {
+        "httpHeaderAuth": {
+          "id": "TBPXSWOF63k9mvm8",
+          "name": "ntfy-token"
+        },
+        "httpBearerAuth": {
+          "id": "nCv4CUN7Ef086Ewj",
+          "name": "Bearer Auth account"
+        }
+      }
+    }
+  ],
+  "pinData": {},
+  "connections": {
+    "Webhook": {
+      "main": [
+        [
+          {
+            "node": "Gate push",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Gate push": {
+      "main": [
+        [
+          {
+            "node": "Power Manager - ensure-on",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Power Manager - ensure-on": {
+      "main": [
+        [
+          {
+            "node": "SSH: pending-raw",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "SSH: pending-raw": {
+      "main": [
+        [
+          {
+            "node": "Split raw files",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Split raw files": {
+      "main": [
+        [
+          {
+            "node": "Nome valido?",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Nome valido?": {
+      "main": [
+        [
+          {
+            "node": "Run one ingest",
+            "type": "main",
+            "index": 0
+          }
+        ],
+        [
+          {
+            "node": "Build ntfy badname",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Build ntfy badname": {
+      "main": [
+        [
+          {
+            "node": "ntfy: send",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    }
+  },
+  "active": true,
+  "settings": {
+    "executionOrder": "v1",
+    "binaryMode": "separate",
+    "timeSavedMode": "fixed",
+    "errorWorkflow": "7Vws3gCX3QnjM3oD",
+    "callerPolicy": "workflowsFromSameOwner",
+    "availableInMCP": false
+  },
+  "versionId": "63863925-606f-4200-824c-52f1919f2bb1",
+  "meta": {
+    "instanceId": "96b2f0ec76a4400bbd481c617b24b3b87024cc7a913efacccaf9fc85722e7417"
+  },
+  "id": "mUJUuQxcDiiPWcUE",
+  "tags": []
+}

deploy/n8n/genome-on-error.json

@@ -0,0 +1,128 @@
+{
+  "name": "Genome: on-error",
+  "nodes": [
+    {
+      "parameters": {},
+      "id": "f715ed51-95e6-475f-8aa5-d0df531cc7cf",
+      "name": "Error Trigger",
+      "type": "n8n-nodes-base.errorTrigger",
+      "typeVersion": 1,
+      "position": [
+        688,
+        -32
+      ]
+    },
+    {
+      "parameters": {
+        "mode": "runOnceForEachItem",
+        "jsCode": "// Global error handler: set this workflow as the \"Error Workflow\" in each genome workflow's\n// Settings. Catches ANY node failure (SSH down, Forgejo 4xx/5xx, etc.) and notifies once.\n// Run Once for Each Item: $json is the error trigger payload.\nconst e = $json.execution || {};\nconst w = $json.workflow || {};\n\n// Safely extract error message from various shapes\nconst rawMsg = (e.error && (e.error.message || e.error.description)) || 'errore sconosciuto';\nconst msg = String(rawMsg).trim();\n\nconst lastNode = e.lastNodeExecuted ? ` (nodo: ${e.lastNodeExecuted})` : '';\nconst workflowName = w.name || 'n8n';\nconst executionUrl = e.url || '';\n\n// Escape markdown to avoid breaking the notification body\nconst msgEsc = msg.replace(/`/g, '\\`').replace(/\\n/g, '\\n');\n\nreturn {\n  topic: 'genome-ingest',\n  title: `Workflow KO · ${workflowName}`,\n  priority: 'high',\n  tags: 'rotating_light',\n  click: executionUrl,\n  actions: executionUrl ? `view, Apri l'esecuzione, ${executionUrl}` : '',\n  body: `**${workflowName}** è fallito${lastNode}.\\n\\n${msgEsc}`\n};"
+      },
+      "id": "dd39bc0f-918a-4645-8f04-540ac9089311",
+      "name": "Build ntfy",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [
+        928,
+        -32
+      ]
+    },
+    {
+      "parameters": {
+        "method": "POST",
+        "url": "=http://ntfy/{{ $json.topic }}",
+        "authentication": "genericCredentialType",
+        "genericAuthType": "httpBearerAuth",
+        "sendHeaders": true,
+        "headerParameters": {
+          "parameters": [
+            {
+              "name": "Title",
+              "value": "={{ $json.title }}"
+            },
+            {
+              "name": "Priority",
+              "value": "={{ $json.priority }}"
+            },
+            {
+              "name": "Tags",
+              "value": "={{ $json.tags }}"
+            },
+            {
+              "name": "Click",
+              "value": "={{ $json.click }}"
+            },
+            {
+              "name": "Actions",
+              "value": "={{ $json.actions }}"
+            },
+            {
+              "name": "Markdown",
+              "value": "yes"
+            }
+          ]
+        },
+        "sendBody": true,
+        "contentType": "raw",
+        "rawContentType": "Raw / Text",
+        "body": "={{ $json.body }}",
+        "options": {
+          "timeout": 15000
+        }
+      },
+      "id": "a9ee90f3-d7fe-445d-96af-12caef46473f",
+      "name": "ntfy: send",
+      "type": "n8n-nodes-base.httpRequest",
+      "typeVersion": 4.4,
+      "position": [
+        1152,
+        -32
+      ],
+      "credentials": {
+        "httpHeaderAuth": {
+          "id": "TBPXSWOF63k9mvm8",
+          "name": "ntfy-token"
+        },
+        "httpBearerAuth": {
+          "id": "nCv4CUN7Ef086Ewj",
+          "name": "Bearer Auth account"
+        }
+      }
+    }
+  ],
+  "pinData": {},
+  "connections": {
+    "Error Trigger": {
+      "main": [
+        [
+          {
+            "node": "Build ntfy",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Build ntfy": {
+      "main": [
+        [
+          {
+            "node": "ntfy: send",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    }
+  },
+  "active": true,
+  "settings": {
+    "executionOrder": "v1",
+    "binaryMode": "separate"
+  },
+  "versionId": "036161c9-c934-474e-9b4f-634259f2a866",
+  "meta": {
+    "instanceId": "96b2f0ec76a4400bbd481c617b24b3b87024cc7a913efacccaf9fc85722e7417"
+  },
+  "id": "7Vws3gCX3QnjM3oD",
+  "tags": []
+}

deploy/n8n/genome-prune.json

@@ -0,0 +1,326 @@
+{
+  "name": "Genome: prune",
+  "nodes": [
+    {
+      "parameters": {
+        "httpMethod": "POST",
+        "path": "forgejo-push-prune",
+        "options": {}
+      },
+      "id": "d31388b9-c6d6-4f28-9a6c-b381922bf5e0",
+      "name": "Webhook prune",
+      "type": "n8n-nodes-base.webhook",
+      "typeVersion": 2.1,
+      "position": [
+        1232,
+        -64
+      ],
+      "webhookId": "d6ac11900058434e"
+    },
+    {
+      "parameters": {
+        "jsCode": "// Gate: proceed ONLY on develop pushes that REMOVED at least one file under raw/.\n// Additions/modifications are handled by the ingest flow; this flow reacts to deletions only.\nconst item = $input.first().json;\nconst b = item.body || item;\nconst ref = String(b.ref || '');\nconst genome = String((b.repository?.name) || '').toLowerCase().trim();\n\n// Branch filter\nif (ref !== 'refs/heads/develop') return [];\n\n// Genome name validation (DNS-like: lowercase alphanum + hyphen, 1-64 chars)\nif (!/^[a-z0-9][a-z0-9-]{0,63}$/.test(genome)) return [];\n\n// Collect removed paths safely\nconst removed = [];\nfor (const c of (b.commits || [])) {\n  if (!c || !Array.isArray(c.removed)) continue;\n  for (const p of c.removed) {\n    if (typeof p === 'string' && p.startsWith('raw/')) {\n      removed.push(p);\n    }\n  }\n}\n\n// Gate: stop if nothing under raw/ was removed\nif (removed.length === 0) return [];\n\nreturn [{ json: { genome, removedCount: removed.length } }];"
+      },
+      "id": "84848a31-d099-459e-bd03-67abc2cf2b77",
+      "name": "Gate prune",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [
+        1456,
+        -64
+      ]
+    },
+    {
+      "parameters": {
+        "workflowId": {
+          "__rl": true,
+          "value": "zbtRXWsLt56nEIfz",
+          "mode": "list",
+          "cachedResultUrl": "/workflow/zbtRXWsLt56nEIfz",
+          "cachedResultName": "Power Manager"
+        },
+        "workflowInputs": {
+          "mappingMode": "defineBelow",
+          "value": {
+            "mode": "ensure-on"
+          },
+          "matchingColumns": [
+            "mode"
+          ],
+          "schema": [
+            {
+              "id": "mode",
+              "displayName": "mode",
+              "required": false,
+              "defaultMatch": false,
+              "display": true,
+              "canBeUsedToMatch": true,
+              "type": "string",
+              "removed": false
+            }
+          ],
+          "attemptToConvertTypes": false,
+          "convertFieldsToString": true
+        },
+        "options": {}
+      },
+      "id": "175e4191-eb1b-4e5d-8d82-c39205753152",
+      "name": "Power Manager - ensure-on",
+      "type": "n8n-nodes-base.executeWorkflow",
+      "typeVersion": 1.3,
+      "position": [
+        1680,
+        -64
+      ]
+    },
+    {
+      "parameters": {
+        "authentication": "privateKey",
+        "command": "=ssh vm101 'pi orphan-wiki {{ $('Gate prune').first().json.genome }}'"
+      },
+      "id": "598f20f8-d668-48da-90e3-1bfada3ace92",
+      "name": "SSH: orphan-wiki",
+      "type": "n8n-nodes-base.ssh",
+      "typeVersion": 1,
+      "position": [
+        1904,
+        -64
+      ],
+      "credentials": {
+        "sshPrivateKey": {
+          "id": "GJQjKzte7Hjdfz89",
+          "name": "n8n container -> n8n-runner@nexus"
+        }
+      }
+    },
+    {
+      "parameters": {
+        "jsCode": "// Gate: proceed to prune only if orphan-wiki actually found orphans.\n// run-prune re-derives independently anyway (no detected-vs-pruned race);\n// this gate just avoids taking the lock for nothing.\nconst out = String($input.first().json.stdout || '').trim();\nlet d;\n\ntry {\n  d = JSON.parse(out);\n} catch (e) {\n  // Malformed JSON from orphan-wiki — log and stop\n  return [{ json: { _gate: 'parse-error', raw: out.substring(0, 500) } }];\n}\n\n// Strict validation: d must be object with numeric count > 0\nif (!d || typeof d !== 'object' || typeof d.count !== 'number' || d.count <= 0) {\n  return [];  // 0 orphans or missing count -> stop silently\n}\n\nreturn [{ json: { genome: d.genome, count: d.count } }];"
+      },
+      "id": "3b644d61-26d8-4024-baed-bcb4ad169a6a",
+      "name": "Orfani?",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [
+        2112,
+        -64
+      ]
+    },
+    {
+      "parameters": {
+        "authentication": "privateKey",
+        "command": "=ssh vm101 'pi prune {{ $json.genome }}'"
+      },
+      "id": "a8cae2c2-6f2f-4ef6-add9-287195aa84b5",
+      "name": "SSH: prune",
+      "type": "n8n-nodes-base.ssh",
+      "typeVersion": 1,
+      "position": [
+        2336,
+        -64
+      ],
+      "credentials": {
+        "sshPrivateKey": {
+          "id": "GJQjKzte7Hjdfz89",
+          "name": "n8n container -> n8n-runner@nexus"
+        }
+      }
+    },
+    {
+      "parameters": {
+        "mode": "runOnceForEachItem",
+        "jsCode": "// Extract the last JSON line from SSH stdout (the command may print logs before/after).\n// Run Once for Each Item: $json is the current SSH result item.\nconst out = String($json.stdout || '').trim();\nconst jsonLines = out\n  .split('\\n')\n  .map(l => l.trim())\n  .filter(l => l.startsWith('{') && l.endsWith('}'));\n\nconst line = jsonLines.pop();  // last JSON object line (command prints JSON last)\n\nlet r;\ntry {\n  r = line ? JSON.parse(line) : { status: 'error', reason: 'nessuna riga JSON trovata in stdout' };\n} catch (e) {\n  r = { status: 'error', reason: 'JSON non parsabile', rawLine: line?.substring(0, 1000) };\n}\n\n// Ensure consistent shape for downstream nodes\nreturn {\n  status: r.status || 'error',\n  reason: r.reason || 'errore sconosciuto',\n  count: r.count,\n  pr_url: r.pr_url,\n  genome: r.genome,\n  _raw: line?.substring(0, 500)\n};"
+      },
+      "id": "da1ab42c-32e1-4c4d-82a1-925fcee1a098",
+      "name": "Parse prune",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [
+        2560,
+        -64
+      ]
+    },
+    {
+      "parameters": {
+        "mode": "runOnceForEachItem",
+        "jsCode": "// Build ntfy notification for genome pruning.\n// Run Once for Each Item: $json is the parsed prune result.\nconst d = $json;\nconst genome = d.genome || 'unknown';\n\nlet n;\nif (d.status === 'ok') {\n  const pm = (d.pr_url || '').match(/\\/pulls\\/(\\d+)/);\n  const num = pm ? `#${pm[1]}` : '';\n  n = {\n    topic: 'genome-ingest',\n    title: `${genome} \\u00b7 potatura ${num}`.replace(/\\s+/g, ' ').trim(),\n    priority: 'default',\n    tags: 'broom',\n    click: d.pr_url || '',\n    actions: d.pr_url ? `view, Apri la PR, ${d.pr_url}` : '',\n    body: `${d.count} sorgente/i orfane proposte per la rimozione. **Approva la PR** per potare, oppure chiudila da Forgejo per annullare.`\n  };\n} else {\n  n = {\n    topic: 'genome-ingest',\n    title: `${genome} \\u00b7 errore potatura`.trim(),\n    priority: 'high',\n    tags: 'rotating_light',\n    click: '',\n    actions: '',\n    body: `${d.reason || 'errore sconosciuto durante la potatura'}.`\n  };\n}\n\nreturn n;"
+      },
+      "id": "ebe99407-6038-4f8f-a73f-7dc7b0a011e0",
+      "name": "Build ntfy",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [
+        2784,
+        -64
+      ]
+    },
+    {
+      "parameters": {
+        "method": "POST",
+        "url": "=http://ntfy/{{ $json.topic }}",
+        "authentication": "genericCredentialType",
+        "genericAuthType": "httpBearerAuth",
+        "sendHeaders": true,
+        "headerParameters": {
+          "parameters": [
+            {
+              "name": "Title",
+              "value": "={{ $json.title }}"
+            },
+            {
+              "name": "Priority",
+              "value": "={{ $json.priority }}"
+            },
+            {
+              "name": "Tags",
+              "value": "={{ $json.tags }}"
+            },
+            {
+              "name": "Click",
+              "value": "={{ $json.click }}"
+            },
+            {
+              "name": "Actions",
+              "value": "={{ $json.actions }}"
+            },
+            {
+              "name": "Markdown",
+              "value": "yes"
+            }
+          ]
+        },
+        "sendBody": true,
+        "contentType": "raw",
+        "rawContentType": "Raw / Text",
+        "body": "={{ $json.body }}",
+        "options": {
+          "timeout": 15000
+        }
+      },
+      "id": "0bd3654e-a73d-4c3a-83ed-9f57ca4aad24",
+      "name": "ntfy: send",
+      "type": "n8n-nodes-base.httpRequest",
+      "typeVersion": 4.4,
+      "position": [
+        2992,
+        -64
+      ],
+      "credentials": {
+        "httpHeaderAuth": {
+          "id": "TBPXSWOF63k9mvm8",
+          "name": "ntfy-token"
+        },
+        "httpBearerAuth": {
+          "id": "nCv4CUN7Ef086Ewj",
+          "name": "Bearer Auth account"
+        }
+      }
+    }
+  ],
+  "pinData": {},
+  "connections": {
+    "Webhook prune": {
+      "main": [
+        [
+          {
+            "node": "Gate prune",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Gate prune": {
+      "main": [
+        [
+          {
+            "node": "Power Manager - ensure-on",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Power Manager - ensure-on": {
+      "main": [
+        [
+          {
+            "node": "SSH: orphan-wiki",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "SSH: orphan-wiki": {
+      "main": [
+        [
+          {
+            "node": "Orfani?",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Orfani?": {
+      "main": [
+        [
+          {
+            "node": "SSH: prune",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "SSH: prune": {
+      "main": [
+        [
+          {
+            "node": "Parse prune",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Parse prune": {
+      "main": [
+        [
+          {
+            "node": "Build ntfy",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Build ntfy": {
+      "main": [
+        [
+          {
+            "node": "ntfy: send",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    }
+  },
+  "active": true,
+  "settings": {
+    "executionOrder": "v1",
+    "binaryMode": "separate",
+    "timeSavedMode": "fixed",
+    "errorWorkflow": "7Vws3gCX3QnjM3oD",
+    "callerPolicy": "workflowsFromSameOwner",
+    "availableInMCP": false
+  },
+  "versionId": "999f640c-aae6-42aa-9a95-aba26987e9d0",
+  "meta": {
+    "instanceId": "96b2f0ec76a4400bbd481c617b24b3b87024cc7a913efacccaf9fc85722e7417"
+  },
+  "id": "smH5Qrv7CQnTtdAF",
+  "tags": []
+}

deploy/n8n/genome-run-one-ingest.json

@@ -0,0 +1,266 @@
+{
+  "name": "Genome: run-one-ingest",
+  "nodes": [
+    {
+      "parameters": {
+        "inputSource": "passthrough"
+      },
+      "id": "b1b7ba8e-1e45-4f76-adc0-089180715975",
+      "name": "On ingest request",
+      "type": "n8n-nodes-base.executeWorkflowTrigger",
+      "typeVersion": 1.1,
+      "position": [
+        224,
+        624
+      ]
+    },
+    {
+      "parameters": {
+        "mode": "runOnceForEachItem",
+        "jsCode": "// SECURITY chokepoint: every ingest to vm101 passes here. Re-validate inputs (defense in depth:\n// callers + the SSH wrapper also validate) and assemble the exact command. Charset-validated\n// fields are safe inside the single-quoted remote command -> no shell injection.\n// Run Once for Each Item: $json is the current ingest request.\nconst d = $json || {};\nconst genome = String(d.genome || '').toLowerCase().trim();\nconst raw    = String(d.raw || '');\nconst mode   = String(d.mode || 'ingest');\nconst fb     = String(d.feedback_b64 || '');\n\nconst okGenome = /^[a-z0-9][a-z0-9-]{0,63}$/.test(genome);\nconst okMode   = (mode === 'ingest' || mode === 'rework');\nconst okRaw    = raw.startsWith('raw/') && !raw.includes('..') && /^[A-Za-z0-9._\\/-]+$/.test(raw);\n// feedback_b64 is required only for rework mode; for ingest it can be empty\nconst okFb     = (mode === 'ingest') || /^[A-Za-z0-9+/=]+$/.test(fb);\n\nif (!okGenome || !okMode || !okRaw || !okFb) {\n  return {\n    _ok: false,\n    genome,\n    mode,\n    _reason: `bad input (genome:${okGenome} mode:${okMode} raw:${okRaw} fb:${okFb})`\n  };\n}\n\n// Build SSH command: single-quoted remote command prevents shell injection\nconst ssh_cmd = (mode === 'rework')\n  ? `ssh vm101 'pi ingest-rework ${genome} ${raw} ${fb}'`\n  : `ssh vm101 'pi ingest ${genome} ${raw}'`;\n\nreturn {\n  _ok: true,\n  ssh_cmd,\n  genome,\n  raw,\n  mode,\n  reason: String(d.reason || ''),\n  prevPr: String(d.prevPr || '')\n};"
+      },
+      "id": "8e538237-0e0e-4308-b2c8-631a52b31185",
+      "name": "Guard & build cmd",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [
+        448,
+        624
+      ]
+    },
+    {
+      "parameters": {
+        "conditions": {
+          "options": {
+            "caseSensitive": true,
+            "typeValidation": "loose",
+            "version": 2
+          },
+          "conditions": [
+            {
+              "id": "4507e3a8b9714c7e",
+              "leftValue": "={{ $json._ok }}",
+              "rightValue": true,
+              "operator": {
+                "type": "boolean",
+                "operation": "true",
+                "singleValue": true
+              }
+            }
+          ],
+          "combinator": "and"
+        },
+        "options": {}
+      },
+      "id": "4b249e76-7ab6-4aa3-886d-06b865931cf6",
+      "name": "Input valido?",
+      "type": "n8n-nodes-base.if",
+      "typeVersion": 2.2,
+      "position": [
+        672,
+        624
+      ]
+    },
+    {
+      "parameters": {
+        "authentication": "privateKey",
+        "command": "={{ $json.ssh_cmd }}"
+      },
+      "id": "8740ae9a-4094-48b2-a9a4-d40d501e09f6",
+      "name": "SSH: ingest",
+      "type": "n8n-nodes-base.ssh",
+      "typeVersion": 1,
+      "position": [
+        880,
+        544
+      ],
+      "credentials": {
+        "sshPrivateKey": {
+          "id": "GJQjKzte7Hjdfz89",
+          "name": "n8n container -> n8n-runner@nexus"
+        }
+      }
+    },
+    {
+      "parameters": {
+        "mode": "runOnceForEachItem",
+        "jsCode": "// run-ingest.sh prints one JSON line; the wrapper may instead print {status:busy|error,...}.\n// Take the last {...} line from stdout (logs may precede/follow).\n// Run Once for Each Item: $json is the current SSH result item.\nconst out = String($json.stdout || '').trim();\nconst jsonLines = out\n  .split('\\n')\n  .map(l => l.trim())\n  .filter(l => l.startsWith('{') && l.endsWith('}'));\n\nconst line = jsonLines.pop();  // last JSON object line (command prints JSON last)\n\nlet r;\ntry {\n  r = line ? JSON.parse(line) : { status: 'error', reason: 'nessuna riga JSON trovata in stdout', raw: out.substring(0, 500) };\n} catch (e) {\n  r = { status: 'error', reason: 'JSON non parsabile', rawLine: line?.substring(0, 1000) };\n}\n\n// Ensure consistent shape for downstream Build ntfy\nreturn {\n  status: r.status || 'error',\n  reason: r.reason || 'errore sconosciuto',\n  pr_url: r.pr_url || '',\n  slug: r.slug || '',\n  lint_clean: r.lint_clean || false,\n  conflict: r.conflict || false,\n  stage: r.stage || '',\n  detail: r.detail || '',\n  log: r.log || '',\n  _raw: line?.substring(0, 500)\n};"
+      },
+      "id": "928344e3-0712-42e0-b1a8-f5caff489746",
+      "name": "Parse result",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [
+        1104,
+        544
+      ]
+    },
+    {
+      "parameters": {
+        "mode": "runOnceForEachItem",
+        "jsCode": "// One builder for ingest + rework outcomes. Title is plain ASCII; the icon comes from Tags\n// (ntfy shortcodes); navigation is via Click (tap) + Actions (button) so it works on every\n// client.\n// Run Once for Each Item: $json is the current parsed result.\n// We read the original request context from the Guard node (same execution, no executeWorkflow in between).\nconst g = $('Guard & build cmd').item.json || {};\nconst verb = (g.mode === 'rework') ? 'rework' : 'ingest';\nconst d = $json || {};\nconst genome = g.genome || 'unknown';\n\n// Build notification based on status\nlet n;\n\nif (g._ok === false) {\n  // Input validation failed (Guard & build cmd rejected it)\n  n = {\n    title: `Errore ${verb}: input non valido`,\n    priority: 'high',\n    tags: 'rotating_light',\n    click: '',\n    actions: '',\n    body: `Richiesta di ${verb} rifiutata.\\n${g._reason || 'motivo sconosciuto'}`\n  };\n} else if (d.status === 'ok') {\n  // Success: PR opened\n  const pm = (d.pr_url || '').match(/\\/pulls\\/(\\d+)/);\n  const num = pm ? `#${pm[1]}` : '';\n  const lint = d.lint_clean ? 'lint pulito' : 'lint con avvisi';\n  const conflict = d.conflict ? ' · ⚠️ conflitto da risolvere' : '';\n  const prevPr = g.prevPr ? ` · sostituisce #${g.prevPr}` : '';\n  const reason = (g.reason && verb === 'ingest') ? ` (${g.reason})` : '';\n\n  n = {\n    title: `${genome} · ${verb} ${d.slug || ''} ${num}`.replace(/\\s+/g, ' ').trim(),\n    priority: d.conflict ? 'high' : 'default',\n    tags: d.conflict ? 'warning' : 'white_check_mark',\n    click: d.pr_url || '',\n    actions: d.pr_url ? `view, Apri la PR, ${d.pr_url}` : '',\n    body: `**${d.slug || 'sorgente'}** ${verb === 'rework' ? 'rilavorata' : 'ingerita'}`\n          + reason + prevPr\n          + `.\\n${lint}${conflict}.`\n  };\n} else if (d.status === 'busy') {\n  // Another ingest is already running on this genome\n  n = {\n    title: `${genome} · ${verb} in coda`,\n    priority: 'min',\n    tags: 'hourglass_flowing_sand',\n    click: '',\n    actions: '',\n    body: `Un altro ingest era in corso su questo genoma. La fonte resta pendente e verrà ripresa al prossimo campanello.`\n  };\n} else if (d.status === 'pr_failed') {\n  // Semantic/lint ok but PR could not be opened\n  const detailLine = String(d.detail || '').split('\\n')[0] || 'dettaglio non disponibile';\n  n = {\n    title: `${genome} · ${d.slug || ''}: PR non aperta`,\n    priority: 'high',\n    tags: 'warning',\n    click: '',\n    actions: '',\n    body: `Semantic e lint ok, ma la PR non si è aperta.\\n${detailLine}`\n  };\n} else {\n  // Generic error (including parse errors)\n  const stage = d.stage ? ` (stage: ${d.stage})` : '';\n  const log = d.log ? `\\nLog: ${d.log}` : '';\n  n = {\n    title: `${genome} · errore ${verb}`,\n    priority: 'high',\n    tags: 'rotating_light',\n    click: '',\n    actions: '',\n    body: `${d.reason || 'errore sconosciuto'}${stage}.${log}`\n  };\n}\n\nn.topic = 'genome-ingest';\nreturn n;"
+      },
+      "id": "9062dfba-02ba-4abc-8be6-828c0b353114",
+      "name": "Build ntfy",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [
+        1328,
+        624
+      ]
+    },
+    {
+      "parameters": {
+        "method": "POST",
+        "url": "=http://ntfy/{{ $json.topic }}",
+        "authentication": "genericCredentialType",
+        "genericAuthType": "httpBearerAuth",
+        "sendHeaders": true,
+        "headerParameters": {
+          "parameters": [
+            {
+              "name": "Title",
+              "value": "={{ $json.title }}"
+            },
+            {
+              "name": "Priority",
+              "value": "={{ $json.priority }}"
+            },
+            {
+              "name": "Tags",
+              "value": "={{ $json.tags }}"
+            },
+            {
+              "name": "Click",
+              "value": "={{ $json.click }}"
+            },
+            {
+              "name": "Actions",
+              "value": "={{ $json.actions }}"
+            },
+            {
+              "name": "Markdown",
+              "value": "yes"
+            }
+          ]
+        },
+        "sendBody": true,
+        "contentType": "raw",
+        "rawContentType": "Raw / Text",
+        "body": "={{ $json.body }}",
+        "options": {
+          "timeout": 15000
+        }
+      },
+      "id": "0c2b4d9b-2700-4815-b47c-8523bc4eb2ff",
+      "name": "ntfy: send",
+      "type": "n8n-nodes-base.httpRequest",
+      "typeVersion": 4.4,
+      "position": [
+        1552,
+        624
+      ],
+      "credentials": {
+        "httpHeaderAuth": {
+          "id": "TBPXSWOF63k9mvm8",
+          "name": "ntfy-token"
+        },
+        "httpBearerAuth": {
+          "id": "nCv4CUN7Ef086Ewj",
+          "name": "Bearer Auth account"
+        }
+      }
+    }
+  ],
+  "pinData": {},
+  "connections": {
+    "On ingest request": {
+      "main": [
+        [
+          {
+            "node": "Guard & build cmd",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Guard & build cmd": {
+      "main": [
+        [
+          {
+            "node": "Input valido?",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Input valido?": {
+      "main": [
+        [
+          {
+            "node": "SSH: ingest",
+            "type": "main",
+            "index": 0
+          }
+        ],
+        [
+          {
+            "node": "Build ntfy",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "SSH: ingest": {
+      "main": [
+        [
+          {
+            "node": "Parse result",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Parse result": {
+      "main": [
+        [
+          {
+            "node": "Build ntfy",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Build ntfy": {
+      "main": [
+        [
+          {
+            "node": "ntfy: send",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    }
+  },
+  "active": true,
+  "settings": {
+    "executionOrder": "v1",
+    "binaryMode": "separate",
+    "timeSavedMode": "fixed",
+    "errorWorkflow": "7Vws3gCX3QnjM3oD",
+    "callerPolicy": "workflowsFromSameOwner",
+    "availableInMCP": false
+  },
+  "versionId": "fd8c1cf6-c5df-4074-b777-113349e32a03",
+  "meta": {
+    "instanceId": "96b2f0ec76a4400bbd481c617b24b3b87024cc7a913efacccaf9fc85722e7417"
+  },
+  "id": "VIi2ovb5gJxNJLbg",
+  "tags": []
+}

deploy/nexus/README.md

@@ -0,0 +1,81 @@
+# Componenti di Sistema — Gestione Sincronizzazione e Automazione Genoma
+
+Questo modulo contiene gli script di backend che vengono installati sul server `nexus` per gestire il ciclo di vita dei vault locali (scratch di lavoro), l'integrazione con Syncthing e l'autocommit dei file grezzi (`raw/`) provenienti dai dispositivi mobili o desktop (es. Obsidian).
+
+## Architettura dei File di Sistema
+
+Gli script sono progettati per girare in un ambiente multi-utente protetto, dove l'istanza globale di `n8n` (tramite l'utente di sistema `n8n-runner`) pilota le operazioni senza possedere i diritti di lettura/scrittura diretti sui file del genoma o sui segreti di configurazione.
+
+### 1. Posizionamento e Permessi degli Script
+
+I file inclusi in questa cartella devono essere installati sul server di produzione nella directory `/usr/local/bin/` con privilegi di esecuzione globali, ma modificabili solo da `root`.
+
+- **Destinazione:** `/usr/local/bin/`
+- **Proprietario (Owner):** `root:root`
+- **Permessi (Chmod):** `0755` (`-rwxr-xr-x`)
+
+#### Elenco degli Script:
+
+- `ensure-genome-vault`: Script idempotente che inizializza o riallinea il vault locale clonandolo da Forgejo (in loopback) sul branch `develop`, configura gli `.stignore` ed effettua il provisioning automatico della cartella condivisa su Syncthing via API.
+- `genome-askpass`: Helper di autenticazione per Git (`GIT_ASKPASS`). Intercetta le richieste di credenziali di Git durante i cloni e i push HTTP su Forgejo, iniettando l'utente e il token applicativo senza esporli nei log di sistema o negli argomenti dei processi.
+- `genome-raw-commit`: Script di polling periodico invocato da n8n. Isola i file modificati nella cartella `raw/`, interroga Syncthing per capire quale dispositivo (e quindi quale autore umano) ha generato la modifica, crea commit atomici attribuiti al singolo autore e pusha le modifiche su Forgejo (`develop`).
+
+---
+
+## Modello di Sicurezza e Visibilità
+
+Per garantire l'isolamento del sistema operativo, l'infrastruttura si basa su tre livelli di confinamento:
+
+### A. Variabili d'Ambiente Protette (`.env`)
+
+Le credenziali (Token Forgejo, API Key Syncthing) risiedono nella Home dell'utente operativo del servizio (`homelab`) e sono completamente invisibili a n8n e ad altri utenti del sistema.
+
+- **Path:** `/home/homelab/.config/knowledge-genome.env`
+- **Permessi:** `0600` (`-rw-------`), di proprietà esclusiva di `homelab:homelab`.
+
+#### env
+
+Nella cartella `~/.config/knowledge-genome.env`.
+
+```text
+# knowledge-genome.env Configuration Profile
+# Requirements: Must be owned by the service user with 0600 permissions.
+
+# Vault path and operational branch
+GENOME_VAULTS_ROOT=/srv/genome-vaults
+GENOME_BASE=develop
+
+# Forgejo Target Instance
+# Replace 127.0.0.1 with vm101 IP if Forgejo is hosted on the virtual machine
+FORGEJO_HOST=127.0.0.1:3001
+FORGEJO_OWNER=Keru
+FORGEJO_USER=n8n-bot
+FORGEJO_TOKEN="............"
+
+# Git Commit Identity
+COMMITTER_NAME=n8n-bot
+COMMITTER_EMAIL=n8n-bot@homelab
+DEFAULT_AUTHOR_NAME="Matteo Cherubini"
+DEFAULT_AUTHOR_EMAIL=matteo@keruhomelab.com
+
+# Syncthing Target Instance
+# Replace 127.0.0.1 with vm101 IP if Syncthing API is hosted on the virtual machine
+SYNCTHING_URL=http://127.0.0.1:8384
+SYNCTHING_API_KEY="............"
+```
+
+### B. Confine dei Privilegi in Sudoers
+
+L'utente di automazione `n8n-runner` (usato dall'agente SSH di n8n) non ha accesso alla shell e non può invocare comandi arbitrari. Può unicamente chiamare i due script principali impersonando l'utente `homelab` senza l'inserimento della password.
+
+Configurazione da applicare in `/etc/sudoers.d/n8n-genome` (con permessi rigorosi `0440`):
+
+```text
+n8n-runner ALL=(homelab) NOPASSWD: /usr/local/bin/ensure-genome-vault, /usr/local/bin/genome-raw-commit
+```
+
+### C. Directory dei Vault
+
+I dati veri e propri sincronizzati da Syncthing risiedono isolati in `/srv/genome-vaults/`.
+
+- **Proprietario**: homelab:homelab (UID/GID 1000), permettendo la convivenza nativa e fluida tra il demone Syncthing in esecuzione nel container e gli script Git locali.

deploy/nexus/ensure-genome-vault.sh

@@ -0,0 +1,126 @@
+#!/bin/bash
+# ensure-genome-vault <genome> [--status-only]
+#
+# Idempotent, unified command for managing genome vaults.
+# Called by n8n during genome creation and as a safety net mechanism.
+#
+# Operation workflow:
+#   - Vault absent  -> Clone from Forgejo (loopback) + track develop branch
+#   - Vault present -> Realign to origin/develop (treated as a rebuildable scratchpad)
+#   - Post-clone/fetch -> Write raw/.stignore and register/update the Syncthing folder.
+#
+# Source of truth is Forgejo. Vaults are scratch spaces and not backed up directly.
+# All operations run locally via loopback.
+
+set -euo pipefail
+genome="${1:?usage: ensure-genome-vault <genome> [--status-only]}"
+mode="${2:-}"
+
+# Slug validation inside the script to prevent path/URL traversal:
+# Lowercase kebab-case, no '/', '..', or spaces.
+[[ "$genome" =~ ^[a-z0-9][a-z0-9-]{0,63}$ ]] || { echo '{"status":"error","reason":"invalid genome name"}'; exit 1; }
+
+set -a; . "${HOME}/.config/knowledge-genome.env"; set +a
+: "${GENOME_VAULTS_ROOT:=/srv/genome-vaults}"
+: "${GENOME_BASE:=develop}"
+: "${FORGEJO_USER:=n8n-bot}"
+: "${FORGEJO_HOST:=127.0.0.1:3001}"
+: "${FORGEJO_OWNER:=Keru}"
+: "${SYNCTHING_URL:=http://127.0.0.1:8384}"
+
+vault="${GENOME_VAULTS_ROOT}/${genome}"
+fid="${genome}-public"
+clone_url="http://${FORGEJO_USER}@${FORGEJO_HOST}/${FORGEJO_OWNER}/${genome}.git"
+export GIT_ASKPASS=/usr/local/bin/genome-askpass   # Provides the n8n-bot token
+
+mkdir -p "$GENOME_VAULTS_ROOT"
+
+# ── 1. Clone (if missing) or realign (if present) ────────────────────────────
+if [[ ! -d "${vault}/.git" ]]; then
+  [[ "$mode" == "--status-only" ]] && { printf '{"status":"absent","genome":"%s"}\n' "$genome"; exit 0; }
+  git clone -q "$clone_url" "$vault"
+  cd "$vault"
+  if git show-ref --verify --quiet "refs/remotes/origin/${GENOME_BASE}"; then
+    git switch -q -c "$GENOME_BASE" --track "origin/${GENOME_BASE}" 2>/dev/null || git switch -q "$GENOME_BASE"
+  else
+    # develop does not exist on remote yet: create it from current base and publish
+    git switch -q -c "$GENOME_BASE"
+    git push -q "$clone_url" "${GENOME_BASE}:${GENOME_BASE}"
+  fi
+  state="cloned"
+else
+  cd "$vault"
+  if [[ "$mode" == "--status-only" ]]; then
+    printf '{"status":"present","genome":"%s","head":"%s"}\n' "$genome" "$(git rev-parse --short HEAD)"
+    exit 0
+  fi
+  git fetch -q origin
+  if git show-ref --verify --quiet "refs/remotes/origin/${GENOME_BASE}"; then
+    git switch -q "$GENOME_BASE" 2>/dev/null || git switch -q -c "$GENOME_BASE" --track "origin/${GENOME_BASE}"
+    # GUARD: hard reset is allowed ONLY if the working tree is clean.
+    # If Syncthing has already written uncommitted raw files, DO NOT destroy them: soft fast-forward.
+    if [[ -z "$(git status --porcelain -- raw/ 2>/dev/null)" ]]; then
+      git reset -q --hard "origin/${GENOME_BASE}"
+      state="realigned"
+    else
+      git merge -q --ff-only "origin/${GENOME_BASE}" 2>/dev/null || true
+      state="realigned-kept-dirty"
+    fi
+  else
+    git switch -q -c "$GENOME_BASE" 2>/dev/null || true
+    git push -q "$clone_url" "${GENOME_BASE}:${GENOME_BASE}"
+    state="base-created"
+  fi
+fi
+
+# ── 2. raw/.stignore + exclusion from git (infrastructure, not content) ────────────
+mkdir -p "${vault}/raw"
+cat > "${vault}/raw/.stignore" <<'EOF'
+// Knowledge Genome — Syncthing exclusions for raw/
+// NEVER unencrypted private data: git-crypt protects INSIDE the repo, not in Syncthing transit
+private
+// Obsidian / editor noise
+.obsidian
+.trash
+*.tmp
+workspace*.json
+// security
+.git
+EOF
+# .stignore must not be included in genome commits
+grep -qxF 'raw/.stignore' "${vault}/.git/info/exclude" 2>/dev/null \
+  || echo 'raw/.stignore' >> "${vault}/.git/info/exclude"
+
+# Syncthing folder marker: must exist on disk (locally, NOT on Git).
+# Without it, Syncthing refuses to scan (“folder marker missing”).
+mkdir -p "${vault}/raw/.stfolder"
+
+# .stfolder must not be included in genome commits
+grep -qxF 'raw/.stfolder' "${vault}/.git/info/exclude" 2>/dev/null \
+|| echo 'raw/.stfolder' >> "${vault}/.git/info/exclude"
+
+# ── 3. Idempotent Syncthing folder configuration (best-effort, does not block the vault) ────────
+folder_state="skipped(no api key)"
+if [[ -n "${SYNCTHING_API_KEY:-}" ]]; then
+  if curl -fsS -o /dev/null -H "X-API-Key: ${SYNCTHING_API_KEY}" \
+       "${SYNCTHING_URL}/rest/config/folders/${fid}" 2>/dev/null; then
+    folder_state="exists"
+  else
+    body="$(curl -fsS -H "X-API-Key: ${SYNCTHING_API_KEY}" \
+              "${SYNCTHING_URL}/rest/config/defaults/folder" \
+            | jq --arg id "$fid" --arg label "${genome} (raw public)" --arg path "${vault}/raw" \
+                 '.id=$id | .label=$label | .path=$path | .type="sendreceive"
+                  | .fsWatcherEnabled=true | .rescanIntervalS=3600')"
+
+    if curl -fsS -o /dev/null -X PUT \
+         -H "X-API-Key: ${SYNCTHING_API_KEY}" -H "Content-Type: application/json" \
+         -d "$body" "${SYNCTHING_URL}/rest/config/folders/${fid}" 2>/dev/null; then
+      folder_state="created"
+    else
+      folder_state="error(check syncthing api)"
+    fi
+  fi
+fi
+
+printf '{"status":"ok","genome":"%s","vault":"%s","state":"%s","syncthing_folder":"%s"}\n' \
+  "$genome" "$vault" "$state" "$folder_state"

deploy/nexus/genome-askpass.sh

@@ -0,0 +1,19 @@
+#!/bin/bash
+#
+# GIT_ASKPASS helper for Forgejo HTTP authentication.
+# Git invokes this script when it needs a username or password.
+#
+
+set -eu
+
+# Load environment variables
+. "${HOME}/.config/knowledge-genome.env"
+
+case "${1:-}" in
+  *[Uu]sername*)
+    printf '%s\n' "${FORGEJO_USER:-n8n-bot}"
+    ;;
+  *)
+    printf '%s\n' "${FORGEJO_TOKEN:?FORGEJO_TOKEN not set}"
+    ;;
+esac

deploy/nexus/genome-raw-commit.sh

@@ -0,0 +1,155 @@
+#!/bin/bash
+# genome-raw-commit <genome>
+#
+# Commit the raw files that Syncthing has placed in the vault and push them to origin/<base>.
+#   - Committer = n8n-bot (sole pusher); Author = the person who wrote it (Syncthing modifiedBy -> .authors.json)
+#   - One commit per author (single-device => one commit). No-op if there is nothing.
+#   - JSON output built with jq (safe escaping), with a `files` array:
+#     for each raw -> file, author, local_path, local_url (file://), remote_url (Forgejo web).
+set -euo pipefail
+genome="${1:?usage: genome-raw-commit <genome>}"
+
+# Input validation to prevent path or URL traversal inside the script
+[[ "$genome" =~ ^[a-z0-9][a-z0-9-]{0,63}$ ]] || { echo '{"status":"error","reason":"invalid genome name"}'; exit 1; }
+
+set -a; . "${HOME}/.config/knowledge-genome.env"; set +a
+: "${GENOME_VAULTS_ROOT:=/srv/genome-vaults}"
+: "${GENOME_BASE:=develop}"
+: "${FORGEJO_USER:=n8n-bot}"
+: "${FORGEJO_HOST:=127.0.0.1:3001}"
+: "${FORGEJO_OWNER:=Keru}"
+: "${FORGEJO_WEB_BASE:=https://git.keruhomelab.com}"   # human-facing URL for remote links (not the loopback)
+: "${SYNCTHING_URL:=http://127.0.0.1:8384}"
+: "${COMMITTER_NAME:=n8n-bot}"
+: "${COMMITTER_EMAIL:=n8n-bot@homelab}"
+: "${DEFAULT_AUTHOR_NAME:=Unknown}"
+: "${DEFAULT_AUTHOR_EMAIL:=unknown@syncthing}"
+
+vault="${GENOME_VAULTS_ROOT}/${genome}"
+fid="${genome}-public"
+authors_map="${GENOME_VAULTS_ROOT}/.authors.json"
+# GENOME_PUSH_URL is a test seam: defaults to the Forgejo loopback URL in production.
+clone_url="${GENOME_PUSH_URL:-http://${FORGEJO_USER}@${FORGEJO_HOST}/${FORGEJO_OWNER}/${genome}.git}"
+export GIT_ASKPASS=/usr/local/bin/genome-askpass
+
+[[ -d "${vault}/.git" ]] || { printf '{"status":"error","reason":"vault absent","genome":"%s"}\n' "$genome"; exit 1; }
+cd "$vault"
+git config user.name  "$COMMITTER_NAME"
+git config user.email "$COMMITTER_EMAIL"
+git config commit.gpgsign false
+
+grep -qxF 'raw/.stignore' "${vault}/.git/info/exclude" 2>/dev/null || echo 'raw/.stignore' >> "${vault}/.git/info/exclude"
+grep -qxF 'raw/.stfolder' "${vault}/.git/info/exclude" 2>/dev/null || echo 'raw/.stfolder' >> "${vault}/.git/info/exclude"
+
+git add -A -- raw/
+git reset -q -- raw/.stignore raw/.stfolder 2>/dev/null || true
+
+# --- Quiet window: only commit raw files that have STOPPED changing. ----------------
+# While a note is being written (Obsidian autosave -> Syncthing -> here) its mtime stays
+# fresh; we leave it UNSTAGED so a half-written note never triggers an ingest. A file is
+# committed only after it has been still for RAW_QUIET_MINUTES. Deletions (nothing on disk)
+# are stable by definition and pass straight through. Deterministic — no model in the loop.
+quiet_min="${RAW_QUIET_MINUTES:-2}"
+held=0
+while IFS= read -r f; do
+  [[ -z "$f" ]] && continue
+  # Only an existing file can be "hot"; a staged deletion has nothing on disk to settle.
+  if [[ -e "$f" && -n "$(find "$f" -mmin -"$quiet_min" 2>/dev/null)" ]]; then
+    git reset -q -- "$f" 2>/dev/null || true
+    held=$((held+1))
+  fi
+done < <(git diff --cached --name-only -- raw/)
+
+if git diff --cached --quiet; then
+  if [[ "$held" -gt 0 ]]; then
+    printf '{"status":"noop","reason":"raw still settling","genome":"%s","held":%d,"quiet_minutes":%d}\n' \
+      "$genome" "$held" "$quiet_min"
+  else
+    printf '{"status":"noop","genome":"%s"}\n' "$genome"
+  fi
+  exit 0
+fi
+
+resolve_dev() {  # $1 = path relative to the vault (raw/...) -> prints the short device id, or empty
+  [[ -z "${SYNCTHING_API_KEY:-}" ]] && return 0
+  curl -fsS -H "X-API-Key: ${SYNCTHING_API_KEY}" --get "${SYNCTHING_URL}/rest/db/file" \
+    --data-urlencode "folder=${fid}" --data-urlencode "file=${1#raw/}" 2>/dev/null \
+    | jq -r '.local.modifiedBy // empty' 2>/dev/null || true
+}
+author_for_dev() {  # $1 = device id -> prints "name\temail"
+  local dev="$1" name="$DEFAULT_AUTHOR_NAME" email="$DEFAULT_AUTHOR_EMAIL"
+  if [[ -n "$dev" && -f "$authors_map" ]] && jq -e --arg d "$dev" '.[$d]' "$authors_map" >/dev/null 2>&1; then
+    name="$(jq -r --arg d "$dev" '.[$d].name'  "$authors_map")"
+    email="$(jq -r --arg d "$dev" '.[$d].email' "$authors_map")"
+  fi
+  printf '%s\t%s' "$name" "$email"
+}
+
+# Collect per-file (relpath, author) and group by author for committing
+declare -A G_FILES G_NAME G_EMAIL
+declare -a ROWS
+while IFS= read -r f; do
+  [[ -z "$f" ]] && continue
+  dev="$(resolve_dev "$f")"
+  IFS=$'\t' read -r aname aemail <<< "$(author_for_dev "$dev")"
+  ROWS+=("${f}"$'\t'"${aname}")
+  key="${aname} <${aemail}>"
+  G_FILES["$key"]+="${f}"$'\n'
+  G_NAME["$key"]="$aname"; G_EMAIL["$key"]="$aemail"
+done < <(git diff --cached --name-only -- raw/)
+
+ts="$(date +%Y-%m-%dT%H:%M:%S%z)"
+commits=0; summary=""
+for key in "${!G_FILES[@]}"; do
+  mapfile -t files < <(printf '%s' "${G_FILES[$key]}")
+  short="$(printf '%s\n' "${files[@]}" | sed 's#^raw/##' | paste -sd, -)"
+  msg="$(printf 'raw(%s): sync %s\n\nAdded-by: %s\nSource: syncthing-autocommit\nSynced-at: %s\n' \
+        "$genome" "$short" "${G_NAME[$key]}" "$ts")"
+  git commit -q --author="$key" -m "$msg" -- "${files[@]}"
+  commits=$((commits+1))
+  summary="${summary}${summary:+; }${G_NAME[$key]}:${short}"
+done
+
+# Push to origin/<base>. The vault is SCRATCH, so we never do an interactive rebase
+# (which can conflict when the same raw file is edited repeatedly). Strategy:
+#   try a fast-forward push; if origin moved, re-apply our raw changes on top of a
+#   fresh origin/<base> and push again. Deterministic, conflict-free.
+git fetch -q origin
+if ! git push -q "$clone_url" "HEAD:${GENOME_BASE}" 2>/dev/null; then
+  # origin advanced: capture our just-made tree for raw/, realign hard, re-apply, retry once.
+  tmp="$(mktemp -d)"
+  cp -a raw/. "$tmp"/ 2>/dev/null || true
+  git reset -q --hard "origin/${GENOME_BASE}"
+  git clean -q -fd
+  cp -a "$tmp"/. raw/ 2>/dev/null || true
+  rm -rf "$tmp"
+  git add -A -- raw/
+  git reset -q -- raw/.stignore raw/.stfolder 2>/dev/null || true
+  if git diff --cached --quiet; then
+    # our content already matches origin -> nothing to push, report ok-noop-after-realign
+    printf '{"status":"ok","genome":"%s","base":"%s","commits":0,"head":"%s","summary":"already in sync after realign","files":[]}\n' \
+      "$genome" "$GENOME_BASE" "$(git rev-parse --short HEAD)"
+    exit 0
+  fi
+  git commit -q --author="${DEFAULT_AUTHOR_NAME} <${DEFAULT_AUTHOR_EMAIL}>" \
+    -m "raw(${genome}): re-apply after realign" -- raw/ || true
+  git push -q "$clone_url" "HEAD:${GENOME_BASE}" \
+    || { printf '{"status":"error","reason":"push-failed-after-realign","genome":"%s"}\n' "$genome"; exit 1; }
+fi
+head="$(git rev-parse --short HEAD)"
+
+# `files` array: local (file://) and remote (Forgejo web) link for each committed raw
+files_json="$(
+  for row in "${ROWS[@]}"; do
+    IFS=$'\t' read -r rel aname <<< "$row"
+    jq -n --arg file "$rel" --arg author "$aname" \
+          --arg lpath "${vault}/${rel}" \
+          --arg lurl  "file://${vault}/${rel}" \
+          --arg rurl  "${FORGEJO_WEB_BASE}/${FORGEJO_OWNER}/${genome}/src/branch/${GENOME_BASE}/${rel}" \
+          '{file:$file, author:$author, local_path:$lpath, local_url:$lurl, remote_url:$rurl}'
+  done | jq -s '.'
+)"
+
+jq -n --arg genome "$genome" --arg base "$GENOME_BASE" --argjson commits "$commits" \
+      --arg head "$head" --arg summary "$summary" --argjson files "$files_json" \
+      '{status:"ok", genome:$genome, base:$base, commits:$commits, head:$head, summary:$summary, files:$files}'

deploy/vm101/README.md

@@ -0,0 +1,60 @@
+# deploy/vm101
+
+System artifacts deployed to **vm101** (the GPU ingest node). The repo is the
+source of truth; the live copies live in `/usr/local/bin/`. Edit here, then
+`sudo ./install.sh` on vm101 to push changes.
+
+## Contents
+
+- `n8n-pi-wrap` — forced-command wrapper that fronts every n8n→vm101 SSH call.
+- `install.sh` — installs the wrapper(s) into `/usr/local/bin` (idempotent).
+
+## n8n-pi-wrap
+
+The only entry point for the `n8n-runner` identity onto vm101. n8n never gets a
+shell here: whatever it sends arrives as `SSH_ORIGINAL_COMMAND`, and a `case`
+whitelist decides what runs. Anything outside the whitelist is denied and logged.
+
+Allowed commands:
+
+| Command | What it does |
+|---|---|
+| `pi run` | one-shot prompt via stdin (proof-of-life / health) |
+| `pi ingest <genome> <raw_path>` | the real two-phase ingest (below) |
+| `ollama list` / `ollama ps` | model introspection |
+
+### The two-phase ingest
+
+`pi ingest` runs the clean-start + two phases, then stops:
+
+1. **Clean start** — `git fetch && switch <INGEST_BASE> && reset --hard origin/<base>`.
+   Destroys only vm101's *scratch* checkout (never a shared branch, never a
+   force-push) — this determinism is by design.
+2. **Semantic** — `skills/ingest/scripts/ingest-semantic.py <genome> <raw_path>`
+   drives `pi` to WRITE `wiki/*` pages + `.ingest-manifest.json`.
+   NOTE: this is the script, NOT `pi -p "/skill:ingest ..."` (that form makes the
+   model reply in chat and write nothing — the classic "manifest not found" trap).
+3. **Mechanical** — `skills/ingest/scripts/run-ingest.sh <genome>` validates the
+   manifest, then index/log/scoped-lint/commit on `feat/ai-ingest-<slug>` and opens
+   a PR onto `<INGEST_BASE>`. Emits one JSON line `{status,slug,pr_url,...}`.
+
+The PR then waits for the human gate. One raw per session, sequential.
+
+### Input hardening
+
+Both inputs come from `SSH_ORIGINAL_COMMAND`, so both are validated:
+
+- `genome` — kebab lowercase `^[a-z0-9-]+$`.
+- `raw_path` — must be under `raw/`, no `..` traversal, restricted charset
+  `[A-Za-z0-9._/-]`, and the file must exist. Rejected paths return a JSON error.
+
+Config (`INGEST_BASE`, `GENOMES_ROOT`, `INGEST_MODEL`, Forgejo token) is sourced
+from `~/.config/knowledge-genome.env` (0600, owner-only).
+
+## Install / update
+
+```bash
+# on vm101
+cd ~/knowledge-genome-orchestrator/deploy/vm101
+sudo ./install.sh
+```

deploy/vm101/install.sh

@@ -0,0 +1,8 @@
+#!/bin/bash
+# deploy/vm101/install.sh — install vm101 wrappers from repo -> /usr/local/bin (idempotent).
+# Run ON vm101 with sudo:  sudo ./install.sh
+set -euo pipefail
+here="$(cd "$(dirname "$0")" && pwd)"
+install -m 0755 "${here}/n8n-pi-wrap" /usr/local/bin/n8n-pi-wrap
+echo "installed: /usr/local/bin/n8n-pi-wrap"
+bash -n /usr/local/bin/n8n-pi-wrap && echo "syntax: ok"

deploy/vm101/n8n-pi-wrap

@@ -0,0 +1,196 @@
+#!/bin/bash
+set -eu
+cmd="${SSH_ORIGINAL_COMMAND:-}"
+case "$cmd" in
+  "pi pending-raw "*)
+    genome="${cmd#pi pending-raw }"
+    case "$genome" in ""|*[!a-z0-9-]*) echo '{"status":"error","reason":"invalid genome name"}'; exit 1;; esac
+    logger -t n8n-pi-wrap "ok: pi pending-raw ${genome}"
+    set -a; . "${HOME}/.config/knowledge-genome.env" 2>/dev/null || true; set +a
+    # Run from the DEPLOYED skill dir (same place as ingest-semantic.py / run-ingest.sh on
+    # lines 54/59), so pending-raw.sh resolves its sibling slug.sh via BASH_SOURCE.
+    exec "${HOME}/.pi/agent/skills/ingest/scripts/pending-raw.sh" "$genome"
+    ;;
+  "pi orphan-wiki "*)
+    genome="${cmd#pi orphan-wiki }"
+    case "$genome" in ""|*[!a-z0-9-]*) echo '{"status":"error","reason":"invalid genome name"}'; exit 1;; esac
+    logger -t n8n-pi-wrap "ok: pi orphan-wiki ${genome}"
+    set -a; . "${HOME}/.config/knowledge-genome.env" 2>/dev/null || true; set +a
+    exec "${HOME}/.pi/agent/skills/ingest/scripts/orphan-wiki.sh" "$genome"
+    ;;
+  "pi run")
+   logger -t n8n-pi-wrap "ok: pi run (prompt via stdin)"
+   prompt=$(cat)
+   exec /usr/local/bin/pi --no-tools --mode json -p "$prompt" </dev/null
+   ;;
+  "pi ingest "*)
+   # Strict positional parse: EXACTLY `pi ingest <genome> <raw_path>` (two tokens).
+   rest="${cmd#pi ingest }"
+   genome="${rest%% *}"
+   raw_path="${rest#* }"
+   # reject: missing second token, or any extra token (a space left in raw_path)
+   if [ "$genome" = "$rest" ] || [ -z "$raw_path" ] || [ "$raw_path" != "${raw_path#* }" ]; then
+     echo '{"status":"error","reason":"usage: pi ingest <genome> <raw_path>"}'; exit 1
+   fi
+   # genome slug: kebab lowercase only
+   case "$genome" in ""|*[!a-z0-9-]*) echo '{"status":"error","reason":"invalid genome name"}'; exit 1;; esac
+   # raw_path whitelist: MUST live under raw/, no traversal, restricted charset.
+   #   - must start with "raw/"      - no ".." segment      - no absolute path / leading slash
+   #   - allowed chars: [A-Za-z0-9._/-]   (kebab slugs + subdirs like raw/articles/foo.md)
+   case "$raw_path" in
+     raw/*) : ;;
+     *) echo '{"status":"error","reason":"raw_path must be under raw/"}'; exit 1;;
+   esac
+   case "$raw_path" in
+     *..*|*//*) echo '{"status":"error","reason":"raw_path traversal"}'; exit 1;;
+   esac
+   case "$raw_path" in
+     *[!A-Za-z0-9._/-]*) echo '{"status":"error","reason":"raw_path illegal chars"}'; exit 1;;
+   esac
+
+   logger -t n8n-pi-wrap "ok: pi ingest ${genome} ${raw_path}"
+
+   # Per-genome lock: serialize writes; never two concurrent ingests on the same genome.
+   exec 9>"/run/lock/kg-ingest-${genome}.lock" 2>/dev/null || exec 9>"/tmp/kg-ingest-${genome}.lock"
+   if ! flock -n 9; then
+     echo '{"status":"busy","reason":"another ingest is running for this genome","genome":"'"$genome"'"}'
+     exit 0
+   fi
+
+   set -a; . "${HOME}/.config/knowledge-genome.env"; set +a
+   cd "${GENOMES_ROOT}/${genome}" || { echo '{"status":"error","reason":"unknown genome"}'; exit 1; }
+
+   # The raw file must actually exist under the genome's raw/ dir.
+   [ -f "$raw_path" ] || { echo '{"status":"error","reason":"raw file not found"}'; exit 1; }
+
+   # Clean start on the configured base (single source of truth in lib/clean-start.sh).
+   : "${KG_LIB_DIR:=${HOME}/knowledge-genome-orchestrator/lib}"
+   source "${KG_LIB_DIR}/clean-start.sh" 2>/dev/null \
+     || { echo '{"status":"error","reason":"clean-start.sh not found"}'; exit 1; }
+   clean_start || { echo '{"status":"error","reason":"clean-start failed"}'; exit 1; }
+
+   # SEMANTIC step: dedicated script drives pi to WRITE wiki pages + manifest.
+   # (NOT `pi -p "/skill:ingest ..."`, which makes the model reply in chat and write nothing.)
+   log="$(mktemp -t pi-ingest.XXXXXX.log)"
+   "${HOME}/.pi/agent/skills/ingest/scripts/ingest-semantic.py" "${genome}" "${raw_path}" \
+       >"$log" 2>&1 \
+     || { echo "{\"status\":\"error\",\"stage\":\"semantic\",\"reason\":\"ingest-semantic failed\",\"log\":\"${log}\"}"; exit 1; }
+
+   # MECHANICAL step: validate manifest -> index/log/scoped-lint/commit/PR -> 1 JSON line
+   exec "${HOME}/.pi/agent/skills/ingest/scripts/run-ingest.sh" "${genome}"
+   ;;
+  "pi prune "*)
+    # Pota le source orfane. Stesso lock dell'ingest (serializza le scritture per genoma),
+    # clean_start, poi run-prune.sh (che ri-deriva gli orfani e apre una PR gated).
+    genome="${cmd#pi prune }"
+    case "$genome" in ""|*[!a-z0-9-]*) echo '{"status":"error","reason":"invalid genome name"}'; exit 1;; esac
+    logger -t n8n-pi-wrap "ok: pi prune ${genome}"
+
+    exec 9>"/run/lock/kg-ingest-${genome}.lock" 2>/dev/null || exec 9>"/tmp/kg-ingest-${genome}.lock"
+    if ! flock -n 9; then
+      echo '{"status":"busy","reason":"another ingest/prune is running for this genome","genome":"'"$genome"'"}'
+      exit 0
+    fi
+
+    set -a; . "${HOME}/.config/knowledge-genome.env"; set +a
+    cd "${GENOMES_ROOT}/${genome}" || { echo '{"status":"error","reason":"unknown genome"}'; exit 1; }
+
+    : "${KG_LIB_DIR:=${HOME}/knowledge-genome-orchestrator/lib}"
+    source "${KG_LIB_DIR}/clean-start.sh" 2>/dev/null \
+      || { echo '{"status":"error","reason":"clean-start.sh not found"}'; exit 1; }
+    clean_start || { echo '{"status":"error","reason":"clean-start failed"}'; exit 1; }
+
+    exec "${HOME}/.pi/agent/skills/ingest/scripts/run-prune.sh" "${genome}"
+    ;;
+  "pi ingest-rework "*)
+    # args: <genome> <raw_path> <feedback_base64>  (3 token).
+    # Feedback in base64 nell'argv: il nodo SSH di n8n non passa stdin, e cosi' i metacaratteri
+    # della review (apici, newline, $(...)) sono neutralizzati.
+    args="${cmd#pi ingest-rework }"
+    genome="${args%% *}"; tmp="${args#* }"
+    raw_path="${tmp%% *}"; fb_b64="${tmp#* }"
+    if [ "$genome" = "$args" ] || [ "$raw_path" = "$tmp" ] || [ -z "$fb_b64" ]; then
+      echo '{"status":"error","reason":"usage: pi ingest-rework <genome> <raw_path> <feedback_b64>"}'; exit 1
+    fi
+    case "$genome"   in ""|*[!a-z0-9-]*)     echo '{"status":"error","reason":"invalid genome"}'; exit 1;; esac
+    case "$raw_path" in raw/*) : ;; *)       echo '{"status":"error","reason":"raw_path must be under raw/"}'; exit 1;; esac
+    case "$raw_path" in *..*|*//*)           echo '{"status":"error","reason":"raw_path traversal"}'; exit 1;; esac
+    case "$raw_path" in *[!A-Za-z0-9._/-]*)  echo '{"status":"error","reason":"raw_path illegal chars"}'; exit 1;; esac
+    case "$fb_b64"   in *[!A-Za-z0-9+/=]*)   echo '{"status":"error","reason":"feedback not base64"}'; exit 1;; esac
+
+    logger -t n8n-pi-wrap "ok: pi ingest-rework ${genome} ${raw_path}"
+    feedback="$(printf '%s' "$fb_b64" | base64 -d 2>/dev/null || true)"
+
+    # lock per-genoma: serializza con gli ingest normali
+    exec 9>"/run/lock/kg-ingest-${genome}.lock" 2>/dev/null || exec 9>"/tmp/kg-ingest-${genome}.lock"
+    if ! flock -n 9; then
+      echo '{"status":"busy","reason":"another ingest is running for this genome","genome":"'"$genome"'"}'; exit 0
+    fi
+    set -a; . "${HOME}/.config/knowledge-genome.env"; set +a
+    cd "${GENOMES_ROOT}/${genome}" || { echo '{"status":"error","reason":"unknown genome"}'; exit 1; }
+    [ -f "$raw_path" ] || { echo '{"status":"error","reason":"raw file not found"}'; exit 1; }
+    : "${KG_LIB_DIR:=${HOME}/knowledge-genome-orchestrator/lib}"
+    source "${KG_LIB_DIR}/clean-start.sh" 2>/dev/null \
+      || { echo '{"status":"error","reason":"clean-start.sh not found"}'; exit 1; }
+    clean_start || { echo '{"status":"error","reason":"clean-start failed"}'; exit 1; }
+    log="$(mktemp -t pi-rework.XXXXXX.log)"
+    INGEST_FEEDBACK="$feedback" \
+      "${HOME}/.pi/agent/skills/ingest/scripts/ingest-semantic.py" "${genome}" "${raw_path}" \
+        >"$log" 2>&1 \
+      || { echo "{\"status\":\"error\",\"stage\":\"semantic\",\"reason\":\"rework failed\",\"log\":\"${log}\"}"; exit 1; }
+    exec "${HOME}/.pi/agent/skills/ingest/scripts/run-ingest.sh" "${genome}"
+    ;;
+  "pi changed-raw "*)
+   # List raw/ files changed between two commits, one per line (the webhook payload
+   # does NOT include file lists, so vm101's checkout computes the diff itself).
+   rest="${cmd#pi changed-raw }"
+   genome="${rest%% *}"
+   range="${rest#* }"
+   before="${range%% *}"
+   after="${range#* }"
+   case "$genome" in ""|*[!a-z0-9-]*) echo '{"status":"error","reason":"invalid genome name"}'; exit 1;; esac
+   case "$before$after" in *[!a-f0-9]*|"") echo '{"status":"error","reason":"invalid commit range"}'; exit 1;; esac
+   logger -t n8n-pi-wrap "ok: pi changed-raw ${genome} ${before}..${after}"
+   set -a; . "${HOME}/.config/knowledge-genome.env"; set +a
+   cd "${GENOMES_ROOT}/${genome}" 2>/dev/null || { echo '{"status":"error","reason":"unknown genome"}'; exit 1; }
+   git fetch -q origin
+   # Resolve the diff base robustly:
+   #   - before all-zero (brand-new branch) or unreachable (force-push) -> fall back to after~1
+   #   - if even after~1 is missing (root commit) -> list all raw files in `after`
+   base="$before"
+   case "$before" in *[!0]*) : ;; *) base="" ;; esac          # all-zero -> empty
+   if [ -n "$base" ] && ! git cat-file -e "${base}^{commit}" 2>/dev/null; then base=""; fi
+   if [ -z "$base" ]; then
+     if git cat-file -e "${after}~1^{commit}" 2>/dev/null; then base="${after}~1"; else base=""; fi
+   fi
+   if [ -n "$base" ]; then
+     files="$(git diff --name-only --diff-filter=d "${base}" "${after}" -- raw/ 2>/dev/null \
+              | grep -vE '(^|/)\.st(folder|ignore)' || true)"
+   else
+     # no usable base: enumerate raw files present at `after`
+     files="$(git ls-tree -r --name-only "${after}" -- raw/ 2>/dev/null \
+              | grep -vE '(^|/)\.st(folder|ignore)' || true)"
+   fi
+   # emit a JSON array via jq (safe escaping)
+   printf '%s\n' "$files" | grep -c . >/dev/null 2>&1 || files=""
+   if [ -z "$files" ]; then
+     echo '{"status":"ok","genome":"'"$genome"'","count":0,"files":[]}'
+   else
+     printf '%s\n' "$files" | jq -R . | jq -s \
+       --arg g "$genome" '{status:"ok", genome:$g, count:length, files:.}'
+   fi
+   ;;
+  "ollama list")
+   logger -t n8n-pi-wrap "ok: ollama list"
+   exec /usr/local/bin/ollama list
+   ;;
+  "ollama ps")
+   logger -t n8n-pi-wrap "ok: ollama ps"
+   exec /usr/local/bin/ollama ps
+   ;;
+ *)
+   logger -t n8n-pi-wrap "denied: ${cmd:-<empty>}"
+   echo "unauthorized command" >&2
+   exit 1
+   ;;
+esac

globals.env

@@ -9,7 +9,7 @@ PROVIDER=forgejo
 
 # --- FORGEJO ---
 FORGEJO_URL=https://git.keruhomelab.com
-FORGEJO_USER=keru
+FORGEJO_USER=Keru
 FORGEJO_SSH_PORT=222
 
 # --- GITHUB (used when PROVIDER=github) ---

lib/clean-start.sh

@@ -0,0 +1,18 @@
+#!/usr/bin/env bash
+# =============================================================================
+# lib/clean-start.sh — single source of truth for the pre-session reset.
+# Caller must already be INSIDE the genome checkout.
+# Aligns the working tree to origin/<base>. Never force-pushes a shared branch.
+# Tolerates a missing remote branch (first-setup scenario).
+# NOTE: sourced library — no `set -euo pipefail` (would leak into the caller).
+# =============================================================================
+
+clean_start() {
+  local base="${INGEST_BASE:-main}"
+  git fetch -q origin || return 1
+  git switch -q "$base" 2>/dev/null || git checkout -q -b "$base" || return 1
+  if git ls-remote --exit-code --heads origin "$base" >/dev/null 2>&1; then
+    git reset -q --hard "origin/${base}" || return 1
+  fi
+  git clean -q -fd || return 1
+}

lib/lint.sh

@@ -208,3 +208,111 @@ check_broken_links() {
     fi
   done <<< "$links"
 }
+
+# ---------------------------------------------------------------------------
+# levenshtein <s1> <s2>
+# Classic edit distance via a two-row rolling buffer, so every array subscript
+# is a single integer. The previous implementation used comma subscripts
+# (d[i,j]); in bash arithmetic the comma operator collapses to one dimension,
+# so the table aliased onto itself and returned wrong distances — it could not
+# even score two identical strings as 0. This form is portable to bash 3.2
+# (no associative arrays). Echoes the integer distance.
+# ---------------------------------------------------------------------------
+levenshtein() {
+  local s1="$1" s2="$2"
+  local len1=${#s1} len2=${#s2}
+  (( len1 == 0 )) && { echo "$len2"; return; }
+  (( len2 == 0 )) && { echo "$len1"; return; }
+
+  local -a prev=() curr=()
+  local i j cost del ins sub min
+  for (( j = 0; j <= len2; j++ )); do prev[j]=$j; done
+
+  for (( i = 1; i <= len1; i++ )); do
+    curr[0]=$i
+    for (( j = 1; j <= len2; j++ )); do
+      cost=1
+      [[ "${s1:i-1:1}" == "${s2:j-1:1}" ]] && cost=0
+      del=$(( prev[j] + 1 ))
+      ins=$(( curr[j-1] + 1 ))
+      sub=$(( prev[j-1] + cost ))
+      min=$del
+      (( ins < min )) && min=$ins
+      (( sub < min )) && min=$sub
+      curr[j]=$min
+    done
+    prev=( "${curr[@]}" )
+  done
+
+  echo "${prev[len2]}"
+}
+
+# ---------------------------------------------------------------------------
+# similarity <s1> <s2>
+# Percentage similarity from the edit distance: 100 = identical, 0 = entirely
+# different. Two empty strings are treated as identical (100), so the divide
+# is always guarded.
+# ---------------------------------------------------------------------------
+similarity() {
+  local s1="$1" s2="$2"
+  local maxlen=${#s1}
+  (( ${#s2} > maxlen )) && maxlen=${#s2}
+  (( maxlen == 0 )) && { echo "100"; return; }
+  local dist
+  dist=$(levenshtein "$s1" "$s2")
+  echo $(( 100 - (dist * 100 / maxlen) ))
+}
+
+# ---------------------------------------------------------------------------
+# check_duplicates <manifest>
+# Advisory only: warns when a page created this run has a slug suspiciously
+# close to an entity/concept already listed in wiki/index.md, so a human can
+# merge them in the PR rather than grow two near-identical pages. Never fails
+# the lint (always returns 0), exactly like check_broken_links.
+#
+# The threshold is tunable via KG_DUP_THRESHOLD (default 70). Exact self-matches
+# are skipped: step 1 of run-ingest.sh appends this run's new slugs to the index
+# BEFORE the lint runs, so without the skip every new slug would match itself at
+# 100%. A page that genuinely collides with a pre-existing file is reported by
+# the manifest as 'modified', not 'created', so skipping created==existing pairs
+# can never mask a real collision.
+# ---------------------------------------------------------------------------
+check_duplicates() {
+  local manifest="$1"
+  [[ -f "$manifest" ]] || return 0
+  command -v jq >/dev/null 2>&1 || return 0
+
+  # New leaf slugs from pages created this run.
+  local -a new_slugs=()
+  local slug
+  while IFS= read -r slug; do
+    [[ -n "$slug" ]] && new_slugs+=("$slug")
+  done < <(jq -r '.pages[]? | select(.status=="created") | .path
+                  | split("/")[-1] | sub("\\.md$";"")' "$manifest" 2>/dev/null)
+
+  # Existing entity/concept slugs already catalogued in the index.
+  local -a existing_slugs=()
+  if [[ -f "wiki/index.md" ]]; then
+    local line
+    while IFS= read -r line; do
+      if [[ $line =~ \[\[(entities|concepts)/([a-z0-9-]+)\]\] ]]; then
+        existing_slugs+=("${BASH_REMATCH[2]}")
+      fi
+    done < "wiki/index.md"
+  fi
+
+  (( ${#new_slugs[@]} && ${#existing_slugs[@]} )) || return 0
+
+  local threshold="${KG_DUP_THRESHOLD:-70}"
+  local new exist sim
+  for new in "${new_slugs[@]}"; do
+    for exist in "${existing_slugs[@]}"; do
+      [[ "$new" == "$exist" ]] && continue   # skip exact self-match (see header)
+      sim=$(similarity "$new" "$exist")
+      if (( sim > threshold )); then
+        warn "Possible duplicate: '${new}' ≈ '${exist}' (${sim}% similar) — review in PR"
+      fi
+    done
+  done
+  return 0
+}

lib/scaffold.sh

@@ -64,8 +64,11 @@ scaffold_genome() {
 
 install_precommit_hook() {
   local repo_path="$1"
-  local hook_path="${repo_path}/.git/hooks/pre-commit"
+  local hooks_dir
+  hooks_dir="$(git -C "$repo_path" rev-parse --git-path hooks)"
+  local hook_path="${hooks_dir}/pre-commit"
 
+  mkdir -p "$hooks_dir"
   cp "${TEMPLATES_DIR}/pre-commit.sh" "$hook_path"
   chmod +x "$hook_path"
   success "Pre-commit security hook installed at: $hook_path"

providers/forgejo.sh

@@ -15,6 +15,7 @@ provider_create_repo() {
   local name="$1"
   local desc="$2"
   local private="$3"
+  local auto_init="${4:-false}"   # genomi: true (submodule add esige un branch). master: false (git init locale + push).
 
   local http_code
   http_code=$(curl -s -o /dev/null -w "%{http_code}" \
@@ -25,7 +26,7 @@ provider_create_repo() {
       \"name\": \"${name}\",
       \"description\": \"${desc}\",
       \"private\": ${private},
-      \"auto_init\": false
+      \"auto_init\": ${auto_init}
     }")
 
   case "$http_code" in

registry.sh

@@ -12,7 +12,7 @@ _REGISTRY_LOADED=1
 PROJECT_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
 
 # Dynamic paths
-WORK_DIR="${HOME}/knowledge-genome-orchestrator"
+WORK_DIR="${PROJECT_ROOT}"
 KEYS_DIR="${WORK_DIR}/keys"
 TEMPLATES_DIR="${PROJECT_ROOT}/templates"
 LIB_DIR="${PROJECT_ROOT}/lib"
@@ -29,5 +29,7 @@ PROVIDERS_DIR="${PROJECT_ROOT}/providers"
 # Example: "genome-work|Work notes and architecture logs||no"
 #          "genome-finance|Personal finance|user/repo-finance|no"
 GENOMES=(
-  "genome-example|Template genome description for knowledge management||no"
+  # Disposable sandbox: exercise the full pipeline (ingest -> PR) end-to-end.
+  # Created by `make setup`. Replace with real domains once the circle is validated.
+  "genome-test|Disposable sandbox for pipeline testing||no"
 )

scripts/setup-genomes.sh

@@ -19,7 +19,7 @@ source "providers/${PROVIDER}.sh"
 step "Processing Genome Registry"
 
 for entry in "${GENOMES[@]}"; do
-    # 3-field format: name|description|linked_repo (linked_repo optional → may be empty)
+    # 4-field format: name|description|linked_repo|cross_source linked_repo optional (may be empty); cross_source defaults to "no".
     IFS='|' read -r GENOME_NAME GENOME_DESC GENOME_LINKED GENOME_CROSS_SOURCE <<< "$entry"
     GENOME_CROSS_SOURCE="${GENOME_CROSS_SOURCE:-no}"
     export GENOME_NAME GENOME_DESC GENOME_LINKED GENOME_CROSS_SOURCE
@@ -27,7 +27,7 @@ for entry in "${GENOMES[@]}"; do
     info "Processing: ${GENOME_NAME} (cross_source: ${GENOME_CROSS_SOURCE})..."
 
     # 1. Remote Creation (Idempotent)
-    provider_create_repo "${GENOME_NAME}" "${GENOME_DESC}" "true"
+    provider_create_repo "${GENOME_NAME}" "${GENOME_DESC}" "true" "true"
     SSH_URL=$(provider_ssh_url "${GENOME_NAME}")
 
     GENOME_PATH="${WORK_DIR}/${MASTER_REPO}/${GENOME_NAME}"
@@ -40,6 +40,8 @@ for entry in "${GENOMES[@]}"; do
 
         cd "${GENOME_NAME}"
 
+        git switch -C main
+
         # IMPORTANT: Initialize git-crypt BEFORE creating any files
         gcrypt_init
 
@@ -50,14 +52,17 @@ for entry in "${GENOMES[@]}"; do
         git add .
         git commit -m "feat: initial scaffold and git-crypt init for ${GENOME_NAME}"
 
-        git branch -M main
-
         git push -u origin main
 
         # Key export and instructions
         gcrypt_export_key "${GENOME_NAME}"
         gcrypt_print_key_instructions "${GENOME_NAME}"
 
+        cd "${WORK_DIR}/${MASTER_REPO}"
+        git add .gitmodules "${GENOME_NAME}"
+        git diff --cached --quiet || git commit -m "chore: register submodule ${GENOME_NAME}"
+        git push
+
         # Commit the submodule reference in the master repo
         cd "${WORK_DIR}/${MASTER_REPO}"
         git commit -m "feat: add ${GENOME_NAME} as submodule"

scripts/verify-genomes.sh

@@ -22,7 +22,7 @@ step "Genome structure: ${MODE}"
 
 TOTAL_MISSING=0
 for entry in "${GENOMES[@]}"; do
-  IFS='|' read -r GENOME_NAME _ _ <<< "$entry"     # 3-field registry; ignore desc + linked
+  IFS='|' read -r GENOME_NAME _ <<< "$entry"       # 4-field registry; only GENOME_NAME used here
   genome_dir="${WORK_DIR}/${MASTER_REPO}/${GENOME_NAME}"
 
   if [[ ! -d "$genome_dir" ]]; then

skills/ingest/SKILL.md

@@ -1,93 +1,92 @@
 ---
 name: ingest
-description: Semantic pass of a single raw source into the current genome's wiki — read the source, write sources/entities/concepts, handle contradictions, then emit a manifest and STOP. Use when a new file lands in raw/. Does NOT do git, log, index, lint, or PRs (a post-processor handles those), and does NOT handle private sources or project repos.
+description: Semantic pass of a single raw source into the current genome's wiki. The model ONLY extracts structured semantic content (summary, entities, concepts, contradictions) and returns one JSON object — it does not write files, produce frontmatter, slugs, git, index, log or PRs. A deterministic conform script (ingest-semantic.py) turns that JSON into properly-structured wiki pages + a manifest; run-ingest.sh then does index/log/lint/PR.
 license: see repository
-compatibility: Runs inside one genome checkout (cwd = genome root). Tools needed — read, edit only. NO bash, NO git. The deterministic steps (index, log, scoped lint, PR) run AFTER you exit, via run-ingest.sh. PRIVATE_CONTEXT must be disabled.
+compatibility: Driven by scripts/ingest-semantic.py (one schema-constrained call to a local model via Ollama /api/chat). NO agent tools are used — no read, no edit, no bash. The model never touches the filesystem. PRIVATE_CONTEXT must be disabled.
-allowed-tools: read edit
 metadata:
   framework: knowledge-genome
   phase: "1-ingest-semantic"
+  mode: structured-json # lightweight agent + deterministic conform
 ---
 
-# Ingest — semantic pass
+# Ingest — semantic pass (structured-JSON)
 
-You run inside ONE genome checkout. `AGENTS.md` (already in your context) is the
+This is the **light** semantic pass. The model's only job is to read one source
-authoritative contract. Your job is the **semantic pass only**: read the source, write
+and return a single JSON object describing what the source contains. It does
-the wiki pages, handle contradictions. You do **not** touch git, the log, the index, the
+**not** write files, choose paths, produce frontmatter, pick slugs, or touch
-linter, or PRs — a post-processor (`run-ingest.sh`) does all of that _after you stop_,
+git / index / log / PRs. All structure is owned by `scripts/ingest-semantic.py`,
-from the manifest you leave behind. This keeps your context clean and your turns few,
+which conforms the model's JSON into wiki pages with enforced kebab-case paths
-which matters on a small local model.
+and frontmatter, and writes `.ingest-manifest.json` in the exact schema
+`run-ingest.sh` consumes. This keeps the agent minimal and makes the output
+impossible to mis-shape, regardless of how small or quirky the local model is.
 
-**Argument:** the relative path of the single raw source to ingest
+Pipeline:
-(e.g. `raw/articles/foo.md`). Process only this one.
 
-## Pre-flight — stop the session if any check fails
+    cd <genome checkout>
+    scripts/ingest-semantic.py <genome> raw/articles/<file>.md   # phase 1 (this)
+    scripts/run-ingest.sh      <genome>                          # phase 2 (deterministic)
 
-1. Refuse if the argument path is under any `private/` directory.
+## Pre-flight (enforced by ingest-semantic.py, not by the model)
+
+1. Refuse if the source path is under any `private/` directory.
 2. Refuse if `PRIVATE_CONTEXT` is not `disabled`.
-3. Confirm the file exists under `raw/`.
+3. Confirm the file exists under `raw/` and is non-empty.
 
-## Semantic work (your only job)
+## What the model returns (the only contract)
 
-1. Read the source once.
+A single JSON object, decoding-constrained to this shape via Ollama's `format`:
-2. Write `wiki/sources/<kebab-slug>.md` — faithful summary + key points, with the required
-   frontmatter (`type: source`, `domain: <genome>`, `maturity: draft`,
-   `last_updated: <today>`, `private: false`, sensible `tags`).
-3. For each entity (person, tool, org) → create or update `wiki/entities/<kebab-name>.md`.
-4. For each concept (pattern, theory, decision) → create or update
-   `wiki/concepts/<kebab-name>.md`.
-5. On a real contradiction with an existing claim, follow `AGENTS.md` §Conflict: create
-   `wiki/queries/conflict-<concept>-<YYYY-MM-DD>.md`. Never overwrite the existing page.
-
-**Naming — you are the sole author of these names; nothing renames your files.** Use
-minimal kebab-case: lowercase letters, digits and hyphens only — no spaces, no underscores,
-no capitals. Pick stable names so the same entity is never created twice (always `acme`,
-never also `acme-corp`). The path you write a file to MUST be byte-for-byte the path you
-list in the manifest.
-
-**Deciding create-vs-update and spotting contradictions — mind the context budget.** Use
-`wiki/index.md` to locate existing pages, then read **only** the handful that _this source
-actually names_ — the entities and concepts in the source's title and opening paragraphs —
-not everything the index lists. When in doubt, read fewer: a missed cross-link is far
-cheaper than a saturated context. Never scan whole directories.
-
-## Finish: write the manifest, then STOP
-
-As your **final action**, write `.ingest-manifest.json` at the genome root
-(NOT under `wiki/`) describing exactly what you did. Then stop — do not commit, lint,
-append to the log/index, or open anything.
 
 ```json
 {
-  "raw_source": "raw/articles/foo.md",
+  "source_title": "Human title of the source",
-  "reasoning": "One sentence for the log: what changed and why.",
+  "source_summary": "Faithful, self-contained prose summary of the source.",
-  "pr_summary": "One or two sentences describing this ingest for the PR.",
+  "key_points": ["Concrete fact or claim worth indexing", "..."],
-  "contradictions": "None   (or: 1 conflict file created — <concept>)",
+  "entities": [
-  "pages": [
+    { "name": "Acme", "kind": "org", "description": "Vendor referenced by the source." }
-    {
+  ],
-      "path": "wiki/sources/foo.md",
+  "concepts": [
-      "summary": "One-line index summary.",
+    { "name": "JWT RS256", "description": "Asymmetric token signing scheme the source uses." }
-      "maturity": "draft",
+  ],
-      "status": "created"
+  "contradictions": [
-    },
+    { "concept": "auth", "description": "Source claims X, contradicting the existing claim Y." }
-    {
+  ],
-      "path": "wiki/entities/acme.md",
+  "reasoning": "One sentence for the log: what this source adds.",
-      "summary": "Acme — vendor.",
+  "pr_summary": "One or two sentences describing this ingest for the PR."
-      "status": "modified"
-    }
-  ]
 }
 ```
 
-Manifest rules:
+Field rules (guidance for the model; the script enforces _structure_):
 
-- List every page you created or modified, with `status` `created` or `modified`.
+- `source_summary` is faithful and in the source's own language. No markdown
-- `summary` is the one-line index description (≈12 words max). For conflict pages the
+  headings inside any description field. No padding.
-  summary is ignored — the index lists conflicts by slug only.
+- `entities` = every person, tool, org or product the source names. `kind` ∈
-- `maturity` is required only on `created` pages (it seeds the new index entry). It is
+  `person|tool|org|product`. `description` = one or two factual sentences.
-  ignored for `modified` pages, so omit it there.
+- `concepts` = every pattern, theory, decision or named idea the source explains.
-- Do NOT add a `model` field — the orchestrator records which model produced this run; you
+- `contradictions` = only a claim that directly contradicts a widely-known fact
-  cannot know your own model name reliably, so do not guess one.
+  or contradicts the source itself; otherwise an empty list.
-- Do not invent a `run_id`, branch, commit, or PR — those belong to the post-processor.
+- Names are the natural name of the thing. The script normalises them to
+  kebab-case and guarantees a single stable page per entity/concept.
 
-One source per session. After writing the manifest, stop.
+## What the conform script guarantees (so the model cannot break it)
+
+- **Paths:** `wiki/sources/<slug>.md`, `wiki/entities/<slug>.md`,
+  `wiki/concepts/<slug>.md`, `wiki/queries/conflict-<concept>-<YYYY-MM-DD>.md`.
+- **Slugs:** minimal kebab-case (lowercase, digits, hyphens; no spaces /
+  underscores / capitals).
+- **Frontmatter:** `type`, `domain: <genome>`, `maturity: draft`,
+  `last_updated: <today>`, `private: false`, `tags`.
+- **Create-vs-update:** existing entity/concept pages are **appended to** (a
+  section attributed to the new source), never overwritten. The source page is
+  the canonical summary of that exact source and is (re)written.
+- **Manifest:** `.ingest-manifest.json` with `raw_source`, `reasoning`,
+  `pr_summary`, `contradictions` (string), and `pages[]` (`path`, `summary`,
+  `status`, plus `maturity` on created pages) — exactly what `run-ingest.sh`
+  validates.
+
+The model name is recorded by the orchestrator (`INGEST_MODEL`); the model does
+not self-report it. No `run_id`, branch, commit or PR is invented here — those
+belong to phase 2.
+
+> Interactive use of `pi` (TUI) is unaffected and still available for manual
+> exploration. The **automated** ingest path no longer relies on `pi` or on
+> native tool-calling: it is the single schema-constrained call above.

skills/ingest/scripts/index-append.py

@@ -1,16 +1,12 @@
 #!/usr/bin/env python3
 # =============================================================================
 # skills/ingest/scripts/index-append.py
-# Insert an entry line into the correct section of wiki/index.md and keep that
+# Insert OR remove an entry line in wiki/index.md, keeping the target section
-# section's entries alphabetically ordered. Bumps frontmatter last_updated.
+# alphabetically ordered. Bumps frontmatter last_updated.
-#
-# NOTE: agents-genome.md and wiki-index.md claim the pre-commit hook sorts the
-# index. The actual pre-commit.sh only runs the plaintext-leak check — it does
-# NOT sort. This script owns the ordering instead. (If you later move sorting
-# into the hook, reduce this to a plain append.)
 #
 #   index-append.py --section Sources \
 #       --entry '- [[sources/foo]] — One-line summary. `maturity: draft`'
+#   index-append.py --remove 'sources/foo'        # delete the entry by wikilink
 # =============================================================================
 import argparse
 import datetime
@@ -22,14 +18,116 @@ LINK_RE = re.compile(r"^- \[\[([^\]]+)\]\]")
 HEADER_RE = re.compile(r"^## ")
 
 
+def bump_last_updated(lines, today):
+    """Bump (or self-heal) last_updated inside the first frontmatter block."""
+    fm_open = False
+    fm_close_idx = None
+    bumped = False
+    for i, ln in enumerate(lines):
+        if ln.strip() == "---":
+            if not fm_open:
+                fm_open = True
+                continue
+            fm_close_idx = i
+            break
+        if fm_open and ln.startswith("last_updated:"):
+            lines[i] = f"last_updated: {today}"
+            bumped = True
+    if not fm_open:
+        print("index-append: warning: no frontmatter found, last_updated not bumped",
+              file=sys.stderr)
+    elif not bumped and fm_close_idx is not None:
+        lines.insert(fm_close_idx, f"last_updated: {today}")
+        print("index-append: last_updated key was missing — inserted", file=sys.stderr)
+
+
+def do_remove(lines, link, today):
+    """Remove every entry line whose wikilink == link. Idempotent."""
+    bump_last_updated(lines, today)
+    kept = []
+    removed = 0
+    for ln in lines:
+        m = LINK_RE.match(ln)
+        if m and m.group(1) == link:
+            removed += 1
+            continue
+        kept.append(ln)
+    if removed:
+        print(f"index-append: removed [[{link}]] ({removed} line(s))")
+    else:
+        # Idempotent: the goal state (entry absent) already holds.
+        print(f"index-append: [[{link}]] not present, nothing to remove")
+    return kept
+
+
+def do_append(lines, section, entry, today):
+    bump_last_updated(lines, today)
+    # Locate the target section [start, end)
+    start = None
+    for i, ln in enumerate(lines):
+        if HEADER_RE.match(ln) and ln[3:].startswith(section):
+            start = i
+            break
+    if start is None:
+        print(f"index-append: section '{section}' not found", file=sys.stderr)
+        return None
+
+    end = len(lines)
+    for i in range(start + 1, len(lines)):
+        if HEADER_RE.match(lines[i]):
+            end = i
+            break
+
+    body = lines[start + 1:end]
+    intro = [ln for ln in body if not ENTRY_RE.match(ln)]
+    entries = [ln for ln in body if ENTRY_RE.match(ln)]
+
+    new_m = LINK_RE.match(entry)
+    new_link = new_m.group(1) if new_m else None
+
+    if new_link is not None:
+        replaced = False
+        for idx, ln in enumerate(entries):
+            m = LINK_RE.match(ln)
+            if m and m.group(1) == new_link:
+                if ln == entry:
+                    print("index-append: entry already present, skipping")
+                    return lines
+                entries[idx] = entry
+                replaced = True
+                break
+        if not replaced:
+            entries.append(entry)
+    else:
+        if entry in entries:
+            print("index-append: entry already present, skipping")
+            return lines
+        entries.append(entry)
+
+    entries.sort(key=str.casefold)
+    while intro and intro[-1].strip() == "":
+        intro.pop()
+    new_section = intro + [""] + entries + [""]
+    print(f"index-append: added to {section}")
+    return lines[:start + 1] + new_section + lines[end:]
+
+
 def main() -> int:
     ap = argparse.ArgumentParser()
-    ap.add_argument("--section", required=True,
+    ap.add_argument("--section", help="Section name (required with --entry)")
-                    help="Section name, e.g. Sources / Entities / Concepts / Queries / Conflicts")
+    ap.add_argument("--entry", help="Full index line to insert")
-    ap.add_argument("--entry", required=True, help="Full index line to insert")
+    ap.add_argument("--remove", metavar="WIKILINK",
+                    help="Remove the entry with this wikilink, e.g. sources/foo")
     ap.add_argument("--file", default="wiki/index.md")
     args = ap.parse_args()
 
+    if bool(args.remove) == bool(args.entry):
+        print("index-append: provide exactly one of --entry or --remove", file=sys.stderr)
+        return 2
+    if args.entry and not args.section:
+        print("index-append: --entry requires --section", file=sys.stderr)
+        return 2
+
     try:
         with open(args.file, encoding="utf-8") as fh:
             lines = fh.read().splitlines()
@@ -38,90 +136,15 @@ def main() -> int:
         return 1
 
     today = datetime.date.today().isoformat()
-
+    if args.remove:
-    # 1. Bump last_updated inside the first frontmatter block
+        out = do_remove(lines, args.remove, today)
-    fm_open = False
-    fm_close_idx = None
-    bumped = False
-    for i, ln in enumerate(lines):
-        if ln.strip() == "---":
-            if not fm_open:
-                fm_open = True
-                continue
-            fm_close_idx = i  # the closing ---
-            break
-        if fm_open and ln.startswith("last_updated:"):
-            lines[i] = f"last_updated: {today}"
-            bumped = True
-
-    if not fm_open:
-        print("index-append: warning: no frontmatter found, last_updated not bumped",
-              file=sys.stderr)
-    elif not bumped and fm_close_idx is not None:
-        # self-heal: frontmatter present but missing the key — insert it before the close
-        lines.insert(fm_close_idx, f"last_updated: {today}")
-        print("index-append: last_updated key was missing — inserted", file=sys.stderr)
-
-    # 2. Locate the target section [start, end)
-    start = None
-    for i, ln in enumerate(lines):
-        if HEADER_RE.match(ln) and ln[3:].startswith(args.section):
-            start = i
-            break
-    if start is None:
-        print(f"index-append: section '{args.section}' not found in {args.file}",
-              file=sys.stderr)
-        return 1
-
-    end = len(lines)
-    for i in range(start + 1, len(lines)):
-        if HEADER_RE.match(lines[i]):
-            end = i
-            break
-
-    # 3. Split the section body into intro (non-entry) and entries
-    body = lines[start + 1:end]
-    intro = [ln for ln in body if not ENTRY_RE.match(ln)]
-    entries = [ln for ln in body if ENTRY_RE.match(ln)]
-
-    # Deduplicate by wikilink PATH, not by exact line: a re-ingest with a changed
-    # summary/maturity should UPDATE the existing entry, not add a duplicate line.
-    new_m = LINK_RE.match(args.entry)
-    new_link = new_m.group(1) if new_m else None
-
-    if new_link is not None:
-        replaced = False
-        for idx, ln in enumerate(entries):
-            m = LINK_RE.match(ln)
-            if m and m.group(1) == new_link:
-                if ln == args.entry:
-                    print("index-append: entry already present, skipping")
-                    return 0
-                entries[idx] = args.entry          # same page, refreshed text
-                replaced = True
-                break
-        if not replaced:
-            entries.append(args.entry)
     else:
-        # No parseable wikilink — fall back to exact-line dedup.
+        out = do_append(lines, args.section, args.entry, today)
-        if args.entry in entries:
+        if out is None:
-            print("index-append: entry already present, skipping")
+            return 1
-            return 0
-        entries.append(args.entry)
-
-    entries.sort(key=str.casefold)
-
-    # Normalise intro: drop trailing blanks, keep header + comment(s)
-    while intro and intro[-1].strip() == "":
-        intro.pop()
-
-    new_section = intro + [""] + entries + [""]
-    lines = lines[:start + 1] + new_section + lines[end:]
 
     with open(args.file, "w", encoding="utf-8") as fh:
-        fh.write("\n".join(lines) + "\n")
+        fh.write("\n".join(out) + "\n")
-
-    print(f"index-append: added to {args.section}")
     return 0
 
 

skills/ingest/scripts/ingest-semantic.py

@@ -0,0 +1,374 @@
+#!/usr/bin/env python3
+# =============================================================================
+# skills/ingest/scripts/ingest-semantic.py
+# Phase 1 (semantic) of the Knowledge Genome ingest — light agent + deterministic conform.
+#
+# - FIXED: Add 'title:' field to frontmatter (lint was complaining about missing title)
+# - NEW: Inject existing index (entity/concept names) into prompt to prevent duplicates
+# - NEW: Richer prompt asking for 2-4 sentences per description (not 1-2), with concrete details
+# - Enhanced schema to handle longer descriptions naturally
+#
+# The model does ONLY semantic extraction and returns ONE schema-constrained JSON
+# object (no tools, no file writing, no git, no frontmatter, no slugs). This script
+# then CONFORMS that output deterministically into wiki pages with enforced
+# frontmatter + kebab-case paths, and writes a .ingest-manifest.json in EXACTLY the
+# schema run-ingest.sh expects.
+#
+#   cd <genome checkout>
+#   ingest-semantic.py <genome> raw/articles/<file>.md      # phase 1 (this)
+#   run-ingest.sh      <genome>                             # phase 2 (deterministic)
+#
+# Emits a single JSON status line on stdout (for n8n / logs).
+# =============================================================================
+import json, os, hashlib, subprocess, re, sys, datetime, urllib.request, urllib.error, time
+
+# --- config (override via env; these live in ~/.config/knowledge-genome.env) ---
+OLLAMA_URL = os.environ.get("OLLAMA_URL", "http://localhost:11434/api/chat")
+MODEL      = os.environ.get("INGEST_MODEL", "qwen2.5:14b")
+NUM_CTX    = int(os.environ.get("INGEST_NUM_CTX", "16384"))
+TIMEOUT    = int(os.environ.get("INGEST_TIMEOUT", "600"))
+# INGEST_THINK: "false" disables a reasoning model's thinking trace, so models like
+# gemma / qwq / qwen3 emit only the structured JSON (no truncation from long thinking).
+# Unset = omit the flag entirely (correct for plain instruct models such as qwen2.5).
+THINK      = os.environ.get("INGEST_THINK")
+TODAY      = datetime.date.today().isoformat()
+FEEDBACK   = os.environ.get("INGEST_FEEDBACK", "").strip()
+
+
+def die(stage, reason):
+    print(json.dumps({"status": "error", "stage": stage, "reason": reason}))
+    sys.exit(1)
+
+
+# --- args + pre-flight (mirror the old skill's guards, enforced in code) ---
+if len(sys.argv) < 3:
+    die("args", "usage: ingest-semantic.py <genome> <raw/rel/path.md>")
+genome  = sys.argv[1]
+raw_rel = sys.argv[2].lstrip("./")
+
+if "private/" in raw_rel or raw_rel.startswith("private"):
+    die("preflight", "refusing private source: " + raw_rel)
+if os.environ.get("PRIVATE_CONTEXT", "disabled") != "disabled":
+    die("preflight", "PRIVATE_CONTEXT must be disabled")
+if not raw_rel.startswith("raw/"):
+    die("preflight", "source must live under raw/: " + raw_rel)
+if not os.path.isfile(raw_rel):
+    die("preflight", "source not found in cwd: " + raw_rel)
+
+with open(raw_rel, "r", encoding="utf-8") as fh:
+    source_text = fh.read()
+if not source_text.strip():
+    die("preflight", "source is empty: " + raw_rel)
+
+# --- pre-flight check: if the prompt exceeds context window, exit cleanly with stage:input ---
+# Conservative estimate: ~4 chars/token for mixed IT/EN text
+SAFETY_MARGIN = 4096  # room for system prompt + JSON response
+MAX_SOURCE_TOKENS = NUM_CTX - SAFETY_MARGIN
+MAX_SOURCE_CHARS = MAX_SOURCE_TOKENS * 4
+
+if len(source_text) > MAX_SOURCE_CHARS:
+    die("input", f"source too large ({len(source_text)} chars, limit ~{MAX_SOURCE_CHARS}). "
+                 f"Use the SPLIT directive or divide the document.")
+
+# --- read existing index to avoid duplicate slugs ---
+existing_entities = set()
+existing_concepts = set()
+if os.path.isfile("wiki/index.md"):
+    try:
+        with open("wiki/index.md", "r", encoding="utf-8") as f:
+            idx_text = f.read()
+        # extract slugs from [[entities/slug]] and [[concepts/slug]] patterns
+        for m in re.finditer(r"\[\[entities/([a-z0-9\-]+)\]\]", idx_text):
+            existing_entities.add(m.group(1))
+        for m in re.finditer(r"\[\[concepts/([a-z0-9\-]+)\]\]", idx_text):
+            existing_concepts.add(m.group(1))
+    except Exception:
+        pass  # index not readable or not found; that's OK
+
+
+def slugify(s):
+    s = re.sub(r"[^a-z0-9]+", "-", (s or "").strip().lower())
+    return re.sub(r"-+", "-", s).strip("-") or "untitled"
+
+
+def twords(s, n=20):
+    """Truncate at n words; used for index entry summaries."""
+    s = " ".join((s or "").split())
+    w = s.split(" ")
+    return s if len(w) <= n else " ".join(w[:n]) + "…"
+
+
+def yaml_dq(s):
+    """Render a value as a YAML double-quoted scalar.
+
+    Titles can contain characters that break a bare scalar — most commonly a
+    colon-space ('Conflict: X' would parse as a mapping), but also '#', leading
+    '-'/'?', quotes, etc. Double-quoting and escaping '\\' and '"' makes any
+    title valid YAML (and keeps Obsidian/Dataview/qmd happy). Newlines are
+    collapsed to spaces so the scalar stays on one line.
+    """
+    s = " ".join((s or "").split())
+    s = s.replace("\\", "\\\\").replace('"', '\\"')
+    return f'"{s}"'
+
+
+def frontmatter(ptype, title, tags):
+    """Return YAML frontmatter with title field."""
+    taglist = "[" + ", ".join(sorted(set(t for t in tags if t))) + "]"
+    return ("---\n"
+            f"title: {yaml_dq(title)}\n"
+            f"type: {ptype}\n"
+            f"domain: {genome}\n"
+            "maturity: draft\n"
+            f"last_updated: {TODAY}\n"
+            "private: false\n"
+            f"tags: {taglist}\n"
+            "---\n")
+
+
+def write_new(path, ptype, title, body, tags):
+    os.makedirs(os.path.dirname(path), exist_ok=True)
+    with open(path, "w", encoding="utf-8") as f:
+        f.write(frontmatter(ptype, title, tags))
+        f.write(f"\n# {title}\n\n{body}\n")
+
+
+def append_section(path, source_slug, body):
+    # never overwrite an existing page: accumulate, attributed to the new source
+    with open(path, "a", encoding="utf-8") as f:
+        f.write(f"\n\n## From [[sources/{source_slug}]]\n\n{body}\n")
+    try:  # best-effort bump of last_updated in the existing frontmatter
+        with open(path, "r", encoding="utf-8") as f:
+            txt = f.read()
+        txt = re.sub(r"(?m)^last_updated:.*$", "last_updated: " + TODAY, txt, count=1)
+        with open(path, "w", encoding="utf-8") as f:
+            f.write(txt)
+    except Exception:
+        pass
+
+
+# --- the semantic contract ---
+SYSTEM_PROMPT = """You perform the SEMANTIC PASS of a single source into a knowledge wiki.
+Read the source and return ONLY structured data describing what it contains.
+You do not write files, you do not produce frontmatter, and you do not invent
+paths, slugs, branches, commits or PRs — a deterministic script does all of that.
+
+Rules:
+- source_summary: a faithful, self-contained summary of the source, in the
+  source's own language. Plain prose, NO markdown headings. 2-4 sentences,
+  with concrete details. Preserve the essence and nuance of the source.
+- key_points: 3-5 concrete facts or claims worth indexing; no padding.
+- entities: every person, tool, organisation or product the source names.
+  kind is one of person|tool|org|product. description is 2-3 factual sentences
+  with specifics. No markdown headings inside the description.
+- concepts: every pattern, theory, decision or named idea the source explains.
+  description is 2-3 factual sentences with concrete examples or context.
+- contradictions: ONLY when the source makes a claim that directly contradicts a
+  widely-known fact or contradicts itself. Otherwise return an empty list.
+- Names must be the natural name of the thing; the script will normalise them.
+
+If the source references an entity or concept already in the wiki (see the list below),
+use the EXACT name already present; do not invent a variant. This prevents duplicates.
+
+Existing entities in this genome:
+{existing_entities}
+
+Existing concepts in this genome:
+{existing_concepts}
+
+Be faithful to the source. Be specific. Do not pad or improvise."""
+
+
+# --- JSON schema -> constrained decoding (Ollama structured outputs) ---
+SCHEMA = {
+    "type": "object",
+    "properties": {
+        "source_title":   {"type": "string"},
+        "source_summary": {"type": "string"},
+        "key_points":     {"type": "array", "items": {"type": "string"}},
+        "entities": {"type": "array", "items": {
+            "type": "object",
+            "properties": {
+                "name":        {"type": "string"},
+                "kind":        {"type": "string",
+                                "enum": ["person", "tool", "org", "product"]},
+                "description": {"type": "string"},
+            },
+            "required": ["name", "description"],
+        }},
+        "concepts": {"type": "array", "items": {
+            "type": "object",
+            "properties": {
+                "name":        {"type": "string"},
+                "description": {"type": "string"},
+            },
+            "required": ["name", "description"],
+        }},
+        "contradictions": {"type": "array", "items": {
+            "type": "object",
+            "properties": {
+                "concept":     {"type": "string"},
+                "description": {"type": "string"},
+            },
+            "required": ["concept", "description"],
+        }},
+        "reasoning":  {"type": "string"},
+        "pr_summary": {"type": "string"},
+    },
+    "required": ["source_title", "source_summary", "entities", "concepts"],
+}
+
+
+def call_model(max_retries=2, base_delay=2.0):
+    """Call Ollama with retry on transient errors (connection, timeout, malformed JSON).
+    Retries up to max_retries times with exponential backoff. Does NOT retry on
+    content errors (schema violations, empty response) — those are model issues."""
+    existing_ents = ", ".join(sorted(existing_entities)) or "(none yet)"
+    existing_conc = ", ".join(sorted(existing_concepts)) or "(none yet)"
+    prompt = SYSTEM_PROMPT.format(existing_entities=existing_ents, existing_concepts=existing_conc)
+
+    user_content = (
+        ("REVISION REQUESTED BY THE MAINTAINER (address this explicitly):\n"
+        + FEEDBACK + "\n\n") if FEEDBACK else ""
+    ) + (
+        "Source path: " + raw_rel + "\n\n--- SOURCE START ---\n"
+        + source_text + "\n--- SOURCE END ---\n\nReturn the JSON now."
+    )
+
+    payload = {
+        "model": MODEL,
+        "messages": [
+            {"role": "system", "content": prompt},
+            {"role": "user", "content": user_content },
+        ],
+        "format": SCHEMA,
+        "stream": False,
+        "options": {"temperature": 0.2, "repeat_penalty": 1.0, "num_ctx": NUM_CTX},
+    }
+    if THINK is not None:
+        payload["think"] = THINK.strip().lower() in ("1", "true", "yes", "on")
+    data = json.dumps(payload).encode("utf-8")
+
+    last_error = None
+    for attempt in range(max_retries + 1):
+        if attempt > 0:
+            delay = base_delay * (2 ** (attempt - 1))
+            print(f"call_model: retry {attempt}/{max_retries} after {delay}s: {last_error}", file=sys.stderr)
+            time.sleep(delay)
+
+        req = urllib.request.Request(OLLAMA_URL, data=data, headers={"Content-Type": "application/json"})
+        try:
+            with urllib.request.urlopen(req, timeout=TIMEOUT) as r:
+                resp = json.loads(r.read().decode("utf-8"))
+        except (urllib.error.URLError, urllib.error.HTTPError, TimeoutError) as e:
+            last_error = f"connection/transport error: {e}"; continue
+        except json.JSONDecodeError as e:
+            last_error = f"invalid JSON from Ollama API: {e}"; continue
+
+        content = ((resp.get("message") or {}).get("content") or "").strip()
+        if content.startswith("```"):
+            content = content.strip("`")
+            brace = content.find("{")
+            if brace >= 0:
+                content = content[brace:]
+        try:
+            return json.loads(content)
+        except json.JSONDecodeError as e:
+            last_error = f"model did not return valid JSON: {e}"
+            if len(content) < 10:
+                continue           # likely truncated -> retry
+            break                  # long but malformed -> model issue, stop
+
+    die("model", last_error or "model call failed after retries")
+
+# --- run the semantic pass ---
+sem = call_model()
+
+# Source of truth: slug from slug.sh --raw (deterministic, path-aware, collision-proof)
+source_slug = subprocess.check_output(
+    ["bash", os.path.join(os.path.dirname(__file__), "slug.sh"), "--raw", raw_rel],
+    text=True
+).strip()
+
+with open(raw_rel, "rb") as f:
+    src_sha = hashlib.sha256(f.read()).hexdigest()
+
+pages = []
+
+# 1. source page — canonical summary of THIS source (re)written
+src_path   = f"wiki/sources/{source_slug}.md"
+src_status = "modified" if os.path.exists(src_path) else "created"
+kp_lines   = "\n".join("- " + p for p in (sem.get("key_points") or []) if p.strip())
+src_body   = (sem.get("source_summary") or "").strip()
+if kp_lines:
+    src_body += "\n\n## Key points\n\n" + kp_lines
+src_body += f"\n\n## Source\n\n- [[{raw_rel}]]\n"
+src_title  = sem.get('source_title') or source_slug
+src_tags   = ([slugify(e.get("name", "")) for e in sem.get("entities", [])]
+              + [slugify(c.get("name", "")) for c in sem.get("concepts", [])])[:8]
+os.makedirs("wiki/sources", exist_ok=True)
+with open(src_path, "w", encoding="utf-8") as f:
+    fm = frontmatter("source", src_title, src_tags)
+    # Inject tracking fields before the closing '---' (first newline-dash-dash-dash-newline)
+    fm = fm.replace("\n---\n", f"\nsource_path: {raw_rel}\nsource_sha256: {src_sha}\n---\n", 1)
+    f.write(fm)
+    f.write(f"\n# {src_title}\n\n{src_body}\n")
+pages.append({"path": src_path,
+              "summary": twords(src_title),
+              "maturity": "draft", "status": src_status})
+
+
+def handle(kind_dir, ptype, items):
+    for it in items or []:
+        name = (it.get("name") or "").strip()
+        if not name:
+            continue
+        slug = slugify(name)
+        path = f"wiki/{kind_dir}/{slug}.md"
+        desc = (it.get("description") or "").strip()
+        if os.path.exists(path):
+            append_section(path, source_slug, desc)
+            pages.append({"path": path, "summary": twords(desc), "status": "modified"})
+        else:
+            body = desc + f"\n\n## Sources\n\n- [[sources/{source_slug}]]\n"
+            write_new(path, ptype, name, body, [genome, ptype])
+            pages.append({"path": path, "summary": twords(desc),
+                          "maturity": "draft", "status": "created"})
+
+
+# 2. entities, 3. concepts
+handle("entities", "entity", sem.get("entities", []))
+handle("concepts", "concept", sem.get("concepts", []))
+
+# 4. contradictions -> conflict pages (run-ingest routes wiki/queries/conflict-*)
+conflicts = sem.get("contradictions") or []
+conf_slugs = []
+for c in conflicts:
+    cslug = slugify(c.get("concept", "unknown"))
+    conf_slugs.append(cslug)
+    path = f"wiki/queries/conflict-{cslug}-{TODAY}.md"
+    write_new(path, "query", f"Conflict: {c.get('concept', '')}",
+              (c.get("description") or "").strip()
+              + f"\n\n## Source\n\n- [[sources/{source_slug}]]\n",
+              [genome, "conflict"])
+    pages.append({"path": path, "summary": "", "maturity": "draft",
+                  "status": "created"})
+
+contradictions_str = ("None" if not conflicts
+                      else f"{len(conflicts)} conflict file(s) created — "
+                           + ", ".join(conf_slugs))
+
+# --- write the manifest in EXACTLY run-ingest.sh's schema ---
+manifest = {
+    "raw_source":     raw_rel,
+    "reasoning":      sem.get("reasoning") or ("Ingest of " + raw_rel),
+    "pr_summary":     sem.get("pr_summary") or ("Semantic ingest of " + raw_rel),
+    "contradictions": contradictions_str,
+    "pages":          pages,
+}
+with open(".ingest-manifest.json", "w", encoding="utf-8") as f:
+    json.dump(manifest, f, indent=2, ensure_ascii=False)
+
+print(json.dumps({"status": "ok", "stage": "semantic",
+                  "pages": len(pages), "model": MODEL,
+                  "manifest": ".ingest-manifest.json"}))

skills/ingest/scripts/log-append.sh

@@ -21,6 +21,7 @@ while [[ $# -gt 0 ]]; do
     --context)   context="$2";   shift 2 ;;
     --output)    output="$2";    shift 2 ;;
     --reasoning) reasoning="$2"; shift 2 ;;
+    --run-id)    run_id_arg="$2"; shift 2 ;;
     *) echo "log-append: unknown arg: $1" >&2; exit 1 ;;
   esac
 done
@@ -35,9 +36,15 @@ esac
 
 [[ -f "$LOG_FILE" ]] || { echo "log-append: not found: $LOG_FILE" >&2; exit 1; }
 
-run_id="$(uuidgen 2>/dev/null || cat /proc/sys/kernel/random/uuid 2>/dev/null || python3 -c 'import uuid; print(uuid.uuid4())')"
+run_id="${run_id_arg:-$(uuidgen 2>/dev/null || cat /proc/sys/kernel/random/uuid 2>/dev/null || python3 -c 'import uuid; print(uuid.uuid4())')}"
 today="$(date +%Y-%m-%d)"
 
+if grep -qF "run_id: \`${run_id}\`" "$LOG_FILE" 2>/dev/null; then
+  echo "log-append: run_id ${run_id} already present — skipping (idempotent)" >&2
+  echo "run_id=${run_id}"
+  exit 0
+fi
+
 {
   printf '\n## [%s] %s | %s\n\n' "$today" "$type" "$subject"
   printf -- '- run_id: `%s`\n'        "$run_id"

skills/ingest/scripts/open-pr.sh

@@ -16,10 +16,11 @@ set -euo pipefail
 : "${FORGEJO_USER:?missing FORGEJO_USER}"
 : "${FORGEJO_TOKEN:?missing FORGEJO_TOKEN}"
 
-slug="" title="" body_file="" base="main" label=""
+slug="" title="" body_file="" base="main" label="" branch=""
 while [[ $# -gt 0 ]]; do
   case "$1" in
     --slug)      slug="$2";      shift 2 ;;
+    --branch)    branch="$2";    shift 2 ;;
     --title)     title="$2";     shift 2 ;;
     --body-file) body_file="$2"; shift 2 ;;
     --base)      base="$2";      shift 2 ;;
@@ -28,16 +29,23 @@ while [[ $# -gt 0 ]]; do
   esac
 done
 
-: "${slug:?--slug required}"
 : "${title:?--title required}"
 : "${body_file:?--body-file required}"
 [[ -f "$body_file" ]] || { echo "open-pr: body file not found: $body_file" >&2; exit 1; }
 
-branch="feat/ai-ingest-${slug}"
+# --branch overrides the default; otherwise derive the ingest branch from --slug.
+# (run-prune passes its own chore/prune-orphans-* branch; run-ingest passes --slug.)
+if [[ -z "$branch" ]]; then
+  : "${slug:?--slug or --branch required}"
+  branch="feat/ai-ingest-${slug}"
+fi
 repo="$(basename -s .git "$(git config --get remote.origin.url)")"
 
 # 1. Branch + commit + push (AGENTS.md rule 5: never commit to main)
-git switch -c "$branch" 2>/dev/null || git switch "$branch"
+# Rolling PR: -C force-resets the branch label to the current base (we are on it after
+# clean_start) and CARRIES the freshly-written wiki/ changes, so a re-ingest of the same
+# source rebuilds the branch cleanly instead of hitting a dirty-switch refusal.
+git switch -C "$branch"
 git add wiki/
 # Scope BOTH the emptiness check and the commit to wiki/ — never commit anything that
 # happened to be staged outside wiki/ (a stray hook, an aborted prior run, etc.).
@@ -46,7 +54,10 @@ if git diff --cached --quiet -- wiki/; then
   exit 1
 fi
 git commit -m "$title" -- wiki/
-git push -u origin "$branch"
+# Try a normal push (new branch / fast-forward). If the branch was rebuilt from base and
+# diverged, force-with-lease updates the open PR in place — the lease refuses to clobber if
+# origin moved unexpectedly since our fetch, so concurrent work is never lost.
+git push -u origin "$branch" 2>/dev/null || git push -u --force-with-lease origin "$branch"
 
 # DRY_RUN: local git work done; skip the Forgejo API (offline tests).
 if [[ -n "${DRY_RUN:-}" ]]; then

skills/ingest/scripts/orphan-wiki.sh

@@ -0,0 +1,35 @@
+#!/usr/bin/env bash
+# =============================================================================
+# orphan-wiki.sh — find source pages whose raw source no longer exists.
+# Reads source_path from each wiki/sources/*.md frontmatter. If the raw is gone,
+# the page is orphaned. Emits JSON envelope: {status, genome, count, files[], detail[]}.
+# Read-only: no lock needed (same policy as pending-raw).
+# =============================================================================
+set -euo pipefail
+
+genome="${1:?usage: orphan-wiki.sh <genome>}"
+base_dir="${GENOMES_ROOT:-${HOME}/genomes}"
+cd "${base_dir}/${genome}" 2>/dev/null || { echo '{"status":"error","reason":"unknown genome"}'; exit 1; }
+
+# Clean start on the configured base (single source of truth in lib/clean-start.sh).
+: "${KG_LIB_DIR:=${HOME}/knowledge-genome-orchestrator/lib}"
+source "${KG_LIB_DIR}/clean-start.sh" 2>/dev/null \
+  || { echo '{"status":"error","reason":"clean-start.sh not found"}'; exit 1; }
+clean_start || { echo '{"status":"error","reason":"clean-start failed"}'; exit 1; }
+
+declare -a ORPH=()
+for page in wiki/sources/*.md; do
+  [[ -e "$page" ]] || continue
+  sp="$(sed -n 's/^source_path:[[:space:]]*//p' "$page" | tr -d '\r' | head -n1)"
+  # Pages without source_path are pre-Step-2 legacy: ignore, don't false-positive.
+  [[ -n "$sp" ]] || continue
+  [[ -f "$sp" ]] || ORPH+=("$page")
+done
+
+if [[ ${#ORPH[@]} -eq 0 ]]; then
+  echo '{"status":"ok","genome":"'"$genome"'","count":0,"files":[],"detail":[]}'
+else
+  for x in "${ORPH[@]}"; do printf '%s\torphan\n' "$x"; done \
+    | jq -R 'split("\t") | {path: .[0], reason: .[1]}' \
+    | jq -s --arg g "$genome" '{status:"ok", genome:$g, count:length, files:[.[].path], detail:.}'
+fi

skills/ingest/scripts/pending-raw.sh

@@ -0,0 +1,64 @@
+#!/usr/bin/env bash
+# =============================================================================
+# pending-raw.sh — deterministic "what needs ingesting" calculator.
+# Reads the clean base checkout and classifies each raw/articles/*.md as:
+#   new       -> no wiki/sources/<slug>.md
+#   modified  -> page exists but its source_sha256 != current file hash
+# Emits the same JSON envelope as changed-raw (drop-in), plus detail[] for ntfy.
+# =============================================================================
+set -euo pipefail
+
+genome="${1:?usage: pending-raw.sh <genome>}"
+base_dir="${GENOMES_ROOT:-${HOME}/genomes}"
+cd "${base_dir}/${genome}" 2>/dev/null || { echo '{"status":"error","reason":"unknown genome"}'; exit 1; }
+
+# Clean start on the configured base (single source of truth in lib/clean-start.sh).
+: "${KG_LIB_DIR:=${HOME}/knowledge-genome-orchestrator/lib}"
+source "${KG_LIB_DIR}/clean-start.sh" 2>/dev/null \
+  || { echo '{"status":"error","reason":"clean-start.sh not found"}'; exit 1; }
+clean_start || { echo '{"status":"error","reason":"clean-start failed"}'; exit 1; }
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+SLUG="${SCRIPT_DIR}/slug.sh"
+
+declare -a NEW=()
+declare -a MOD=()
+declare -A SEEN_SLUG=()
+
+if [[ -d raw/articles ]]; then
+  while IFS= read -r -d '' f; do
+    rel="${f#./}"
+    case "$rel" in
+      */.stfolder/*|*/.stignore|*/.gitkeep) continue ;;
+    esac
+    slug="$("$SLUG" --raw "$rel")" || continue
+
+    # Residual collision (two distinct raws -> same slug): warn, do not silence.
+    if [[ -n "${SEEN_SLUG[$slug]:-}" && "${SEEN_SLUG[$slug]}" != "$rel" ]]; then
+      logger -t pending-raw "warn: slug collision ${slug}: ${SEEN_SLUG[$slug]} <-> ${rel}"
+    fi
+    SEEN_SLUG[$slug]="$rel"
+
+    page="wiki/sources/${slug}.md"
+    if [[ ! -f "$page" ]]; then
+      NEW+=("$rel")
+    else
+      cur="$(sha256sum "$rel" | cut -d' ' -f1)"
+      rec="$(sed -n 's/^source_sha256:[[:space:]]*//p' "$page" | tr -d '\r' | head -n1)"
+      if [[ "$cur" != "$rec" ]]; then
+        MOD+=("$rel")
+      fi
+    fi
+  done < <(find raw/articles -type f -name '*.md' -print0 2>/dev/null)
+fi
+
+if [[ ${#NEW[@]} -eq 0 && ${#MOD[@]} -eq 0 ]]; then
+  echo '{"status":"ok","genome":"'"$genome"'","count":0,"files":[],"detail":[]}'
+else
+  {
+    for x in "${NEW[@]}"; do printf '%s\tnew\n' "$x"; done
+    for x in "${MOD[@]}"; do printf '%s\tmodified\n' "$x"; done
+  } | jq -R 'split("\t") | {path: .[0], reason: .[1]}' \
+    | jq -s --arg g "$genome" \
+        '{status: "ok", genome: $g, count: length, files: [.[].path], detail: .}'
+fi

skills/ingest/scripts/run-ingest.sh

@@ -1,13 +1,17 @@
 #!/usr/bin/env bash
 # =============================================================================
 # skills/ingest/scripts/run-ingest.sh
-# Post-pi orchestrator. Runs OUTSIDE pi's loop, on vm101, in the genome checkout.
+# Post-semantic orchestrator. Runs OUTSIDE the model, on vm101, in the genome
-# Consumes .ingest-manifest.json (written by the ingest skill) and performs every
+# checkout. Consumes .ingest-manifest.json (written by ingest-semantic.py) and
-# deterministic step — index, log, scoped lint, PR — so pi's context stays clean.
+# performs every deterministic step — index, log, scoped lint, PR.
 #
 #   run-ingest.sh <genome_name> [manifest_path]
 #
 # Emits a single JSON result line on stdout for n8n to parse.
+#
+# every page listed in the manifest must exist on disk before we trust the run.
+# Everything else is unchanged: the manifest the semantic phase now produces is
+# already in this script's expected schema.
 # =============================================================================
 set -euo pipefail
 
@@ -49,7 +53,7 @@ contradictions="$(jq -r '.contradictions // "None"' "$manifest")"
 
 [[ -n "$raw_source" && "$raw_source" != "null" ]] || fail "manifest" "raw_source missing"
 
-slug="$(bash "${SCRIPTS}/slug.sh" "$raw_source")" || fail "slug" "empty or invalid slug for ${raw_source}"
+slug="$(bash "${SCRIPTS}/slug.sh" --raw "$raw_source")" || fail "slug" "empty or invalid slug for ${raw_source}"
 
 # --- collect touched paths ---
 mapfile -t created_paths  < <(jq -r '.pages[] | select(.status=="created")  | .path' "$manifest")
@@ -57,6 +61,13 @@ mapfile -t modified_paths < <(jq -r '.pages[] | select(.status=="modified") | .p
 all_paths=( "${created_paths[@]}" "${modified_paths[@]}" )
 [[ ${#all_paths[@]} -gt 0 ]] || fail "manifest" "no pages reported"
 
+# --- the semantic phase (ingest-semantic.py) writes the files; verify
+# every manifest page actually exists on disk before trusting the run. Catches any
+# drift between what the manifest claims and what was really written. ---
+for _p in "${all_paths[@]}"; do
+  [[ -f "$_p" ]] || fail "pages" "manifest lists a file not present on disk: ${_p}"
+done
+
 conflict_label=""
 
 # NOTE: No rollback. The steps below modify the working tree in order (index → log → commit).
@@ -96,19 +107,31 @@ done < <(jq -r '.pages[] | select(.status=="created")
                 | [.path, (.summary // ""), (.maturity // "draft")] | @tsv' "$manifest")
 
 # --- 2. log entry ---
+# Stable run_id: deterministic from the input (raw path + content hash). Survives wrapper
+# re-runs and makes the append-only log idempotent (paired with the guard in log-append.sh).
+src_sha="$(sha256sum "$raw_source" 2>/dev/null | cut -d' ' -f1)" || src_sha="unknown"
+run_id="$(printf '%s' "${raw_source}:${src_sha}" | sha256sum | cut -c1-16)"
 out="$(jq -r '[.pages[].path | "[[" + (sub("^wiki/";"") | sub("\\.md$";"")) + "]]"] | join(", ")' "$manifest")"
-bash "${SCRIPTS}/log-append.sh" --type INGEST --subject "$slug" --model "$model" \
+bash "${SCRIPTS}/log-append.sh" --run-id "$run_id" --type INGEST --subject "$slug" --model "$model" \
   --context "[[${raw_source}]]" --output "${out:-*(none)*}" --reasoning "$reasoning" \
   || fail "log" "log-append failed"
 
 # --- 3. scoped linter (capture findings for the PR; never aborts the run) ---
-lint_out="$( bash "${SCRIPTS}/scoped-lint.sh" "$genome" "${all_paths[@]}" 2>&1 )" && lint_rc=0 || lint_rc=$?
+# Point scoped-lint at the same manifest we were handed so its duplicate
+# advisory reads the right file even when a non-default path arrives as $2.
+# (The dedup check lives inside lib/lint.sh and is invoked by scoped-lint —
+# there is no separate check-duplicates.sh script.)
+export INGEST_MANIFEST="$manifest"
+lint_out="$(
+  bash "${SCRIPTS}/scoped-lint.sh" "$genome" "${all_paths[@]}" 2>&1
+)" && lint_rc=0 || lint_rc=$?
 
 # --- 4. assemble the PR body (manifest tables + lint results) ---
 body="$(mktemp)"
 trap 'rm -f "$body"' EXIT  # auto-clean on any exit (success, fail(), or crash)
 
 {
+  echo "<!-- kg:raw=${raw_source} -->" # marker for the rejection loop (invisible in the render)
   echo "## Summary"
   echo "$pr_summary"
   echo ""

skills/ingest/scripts/run-prune.sh

@@ -0,0 +1,96 @@
+#!/usr/bin/env bash
+# =============================================================================
+# skills/ingest/scripts/run-prune.sh
+# Symmetric companion to run-ingest: prune source pages whose raw source no
+# longer exists. RE-DERIVES the orphan set itself (mirrors orphan-wiki.sh) — it
+# never trusts a list handed in by n8n, so there is no "detected-vs-pruned"
+# race. Removes ONLY the pages it derived plus their index entries, commits
+# ONLY wiki/ on chore/prune-orphans-<date>, and opens a GATED removal PR (the
+# operator approves the deletion; principle 2). Never deletes of its own accord.
+#
+# Runs OUTSIDE the model, on vm101, cwd = genome checkout. The wrapper (`pi
+# prune`) has already taken the per-genome lock and done clean_start, exactly
+# like `pi ingest` — so this script does neither.
+#
+#   run-prune.sh <genome>
+#
+# Emits a single JSON result line on stdout for n8n to parse.
+# =============================================================================
+set -euo pipefail
+
+genome="${1:?usage: run-prune.sh <genome>}"
+SCRIPTS="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+
+fail() {
+  jq -nc --arg stage "$1" --arg reason "$2" '{status:"error", stage:$stage, reason:$reason}'
+  exit 1
+}
+
+command -v jq      >/dev/null 2>&1 || { echo '{"status":"error","reason":"jq missing"}'; exit 1; }
+command -v python3 >/dev/null 2>&1 || fail "deps" "python3 missing (needed by index-append.py)"
+
+# --- re-derive orphans (same rule as orphan-wiki.sh; computed fresh, here, now) ---
+# A wiki/sources/*.md page is orphaned when its frontmatter source_path points at
+# a raw file that no longer exists. Legacy pages without source_path are ignored.
+declare -a ORPH=()
+for page in wiki/sources/*.md; do
+  [[ -e "$page" ]] || continue
+  sp="$(sed -n 's/^source_path:[[:space:]]*//p' "$page" | tr -d '\r' | head -n1)"
+  [[ -n "$sp" ]] || continue
+  [[ -f "$sp" ]] || ORPH+=("$page")
+done
+
+if [[ ${#ORPH[@]} -eq 0 ]]; then
+  jq -nc '{status:"ok", count:0, pruned:[], detail:"no orphans"}'
+  exit 0
+fi
+
+# --- remove each orphan page + its index entry (anti-traversal, wiki/-only) ---
+declare -a PRUNED=()
+for page in "${ORPH[@]}"; do
+  case "$page" in
+    wiki/*) : ;;
+    *) fail "prune" "refusing to remove outside wiki/: ${page}" ;;
+  esac
+  case "$page" in *..*) fail "prune" "path traversal in page: ${page}" ;; esac
+  [[ -f "$page" ]] || continue
+  rm -f "$page"
+  link="${page#wiki/}"; link="${link%.md}"            # e.g. sources/foo
+  python3 "${SCRIPTS}/index-append.py" --remove "$link" \
+    || fail "index" "index-append --remove failed for ${link}"
+  PRUNED+=("$link")
+done
+
+# --- assemble the PR body ---
+date_tag="$(date +%F)"
+body="$(mktemp)"
+trap 'rm -f "$body"' EXIT
+{
+  echo "## Prune orphaned sources"
+  echo ""
+  echo "These source pages reference a \`source_path\` whose raw file no longer exists"
+  echo "in \`raw/\`. Removing them keeps the wiki in sync with git (the source of truth)."
+  echo ""
+  echo "| Removed page |"
+  echo "|--------------|"
+  for l in "${PRUNED[@]}"; do echo "| \`wiki/${l}.md\` |"; done
+} > "$body"
+
+# --- open the GATED removal PR on a chore/ branch (open-pr --branch override) ---
+branch="chore/prune-orphans-${date_tag}"
+pr_out="$( bash "${SCRIPTS}/open-pr.sh" \
+  --branch "$branch" \
+  --title "chore: prune ${#PRUNED[@]} orphaned source(s)" \
+  --body-file "$body" --base "${INGEST_BASE:-main}" 2>&1 )" && pr_rc=0 || pr_rc=$?
+pr_url="$(printf '%s\n' "$pr_out" | sed -n 's/^PR opened: //p' | head -n1)"
+
+# --- result line for n8n ---
+jq -nc \
+  --arg status "$([[ $pr_rc -eq 0 ]] && echo ok || echo pr_failed)" \
+  --argjson count "${#PRUNED[@]}" \
+  --arg pr_url "$pr_url" \
+  --arg detail "$pr_out" \
+  --argjson pruned "$(printf '%s\n' "${PRUNED[@]}" | jq -R . | jq -s .)" \
+  '{status:$status, count:$count, pr_url:$pr_url, pruned:$pruned, detail:$detail}'
+
+[[ $pr_rc -eq 0 ]] || exit 1

skills/ingest/scripts/scoped-lint.sh

@@ -49,6 +49,13 @@ for f in "$@"; do
   check_broken_links        "$f" || true   # warnings only
 done
 
+# Cross-page duplicate advisory: runs ONCE over the whole manifest (not per
+# file) — it compares this run's created slugs against the index, so repeating
+# it for every file would only print the same warnings N times. Warn-only;
+# never affects the exit status. INGEST_MANIFEST lets run-ingest.sh point us at
+# a non-default manifest path; falls back to the conventional name otherwise.
+check_duplicates "${INGEST_MANIFEST:-.ingest-manifest.json}"
+
 echo ""
 echo "scoped-lint: ${errors} error(s), ${stale} stale across ${count} file(s)"
 

skills/ingest/scripts/slug.sh

@@ -7,6 +7,18 @@
 # =============================================================================
 set -euo pipefail
 
+if [[ "${1:-}" == "--raw" ]]; then
+  raw="${2:?usage: slug.sh --raw <raw/bucket/rel/path>}"
+  rel="${raw#raw/}"; rel="${rel#*/}"          # strip "raw/" and the bucket name
+  rel="${rel%.*}"                              # strip extension
+  slug="$(printf '%s\n' "$rel" | tr '/' '\n' \
+          | sed -E 's/[^a-zA-Z0-9]+/-/g; s/-{2,}/-/g; s/^-+//; s/-+$//' \
+          | tr '[:upper:]' '[:lower:]' | paste -sd- -)"
+  [[ -n "$slug" ]] || { echo "slug: empty result for input '${raw}'" >&2; exit 1; }
+  printf '%s\n' "$slug"
+  exit 0
+fi
+
 input="${1:?usage: slug.sh <path-or-title>}"
 
 # Strip directory and extension when given a path

templates/agents-genome.md

@@ -166,7 +166,7 @@ private: true | false
 ### Links
 
 - Internal: `[[folder/file]]` — Obsidian wikilinks only. Never `[text](url)` for internal refs.
-- Cross-genome: `[[../genome-target/wiki/folder/file]]`.
+- Cross-genome: NOT via wikilink (submodule pointers make relative paths brittle). A concept owned by another genome is pulled in by the navigation skill as a raw under `raw/articles/crossgen-<topic>-<date>.md`, then ingested here normally. See master `AGENTS.md` §Cross-Genome Pull.
 - External: `[text](https://...)`.
 
 ### Index entries

templates/pr-description.md

@@ -2,6 +2,19 @@
 
 <!-- One sentence: goal of this session and source processed. -->
 
+<!--
+REVIEW GUIDELINES (write the guideline as the FIRST word of your review):
+  REWORK:  <what to fix>     -> same branch, guided retry
+  RESTART: <why restart>    -> close PR, start over from scratch
+  SPLIT:   <how to split>       -> close PR, reopen as separate branches
+  REJECT:  <why not>           -> close PR, no retry
+  MERGE                          -> approve and merge
+Rules: one concern per directive; be specific to lines/pages; name the principle
+that was violated; describe the DESIRED STATE; avoid saying “do better.”
+-->
+
+Translated with DeepL.com (free version)
+
 ## Pages Created
 
 | Path              | Type                              | Maturity |

templates/pre-commit.sh

@@ -11,9 +11,8 @@ set -euo pipefail
 FAILED=0
 
 # Verify git-crypt is initialized
-if [[ ! -d ".git-crypt" ]]; then
+if ! git-crypt status >/dev/null 2>&1; then
-    printf "\n\033[0;31m[CRITICAL] git-crypt not initialized.\033[0m\n"
+    printf "\n[CRITICAL] git-crypt not initialized.\n"
-    printf "Run 'git-crypt init' and 'make setup' before committing.\n"
     exit 1
 fi
 

templates/wiki-log.md

@@ -57,7 +57,7 @@ grep "^## \[2026-05" wiki/log.md
 ## [{{DATE}}] CONFIG | Genome scaffolded
 
 - run_id: `system-init`
-- model: `setup-knowledge-genome.sh`
+- model: `scaffold.sh`
 - context_read: _(none — initial scaffold)_
 - output_written: `[[wiki/index.md]]`, `[[wiki/log.md]]`, `[[AGENTS.md]]`
 - reasoning: Initial directory structure and encryption layer initialized by setup script.

tests/README.md

@@ -9,7 +9,7 @@ they run anywhere (laptop, CI, a git hook). They do **not** belong on vm101 or i
 | File | Covers |
 |------|--------|
 | `scripts.bats`    | `slug.sh`, `log-append.sh`, `index-append.py` (insert, sort, bump, idempotent) |
-| `lint.bats`       | `lib/lint.sh` validators + `scoped-lint.sh` reuse |
+| `lint.bats`       | `lib/lint.sh` validators + `scoped-lint.sh` reuse + duplicate-slug advisory (edit-distance math, self-match skip, once-per-run) |
 | `structure.bats`  | `lib/structure.sh` report/sync |
 | `run-ingest.bats` | `run-ingest.sh` end-to-end (DRY_RUN, local bare remote) — needs `jq` |
 

tests/clean-start.bats

@@ -0,0 +1,18 @@
+#!/usr/bin/env bats
+setup() {
+  load 'helpers'
+  source "${LIB_DIR}/clean-start.sh" 2>/dev/null || source "${REPO_ROOT}/lib/clean-start.sh"
+}
+@test "clean_start: aligns to origin/base, reverts tracked edits, removes untracked" {
+  G="$(make_fixture_genome)"; cd "$G"
+  echo "from origin" >> wiki/index.md
+  git add -A && git commit -q -m "origin ahead" && git push -q
+  git reset --hard HEAD~1                 # local BEHIND origin/main
+  echo "local junk" >> wiki/log.md        # tracked edit, uncommitted
+  echo "scratch" > scratch.txt            # genuinely untracked
+  INGEST_BASE="main" clean_start
+  git diff --quiet origin/main            # aligned to origin
+  grep -q "from origin" wiki/index.md     # forwarded to origin state
+  ! grep -q "local junk" wiki/log.md      # tracked edit reverted
+  [ ! -f scratch.txt ]                    # untracked removed
+}

tests/helpers.bash

@@ -57,8 +57,8 @@ private: false
 
 
 ## Private Synthesis (`wiki/private/`)
-*Restricted access. Requires PRIVATE_CONTEXT: enabled and unlocked repo.*
+_Restricted access. Requires `PRIVATE_CONTEXT: enabled` and unlocked repo._
-*List slug names ONLY. Do not append summaries — prevents metadata leakage.*
+_List slug names ONLY. Do not append summaries — prevents metadata leakage._
 EOF
 
   cat > "${g}/wiki/log.md" <<'EOF'
@@ -88,16 +88,17 @@ EOF
     git init -q
     # Hermetic: ignore the user's global git config (signing, global hooks);
     # otherwise commit.gpgsign or a global core.hooksPath makes git commit fail here.
-    git config commit.gpgsign false
+    git config --local user.name "Framework Test"
-    git config core.hooksPath "${base}/nohooks"
+    git config --local user.email "test@genome.local"
-    git config user.email t@t
+    git config --local commit.gpgsign false
-    git config user.name tester
+    git config --local core.hooksPath "${base}/nohooks"
-    git add .
+
-    git commit -qm init
     git branch -M main
     git remote add origin "${base}/origin.git"
+    git add .
+    git commit -q -m "chore: initial scaffold"
     git push -q -u origin main
-  ) >/dev/null
+  )
 
   echo "${g}"
 }

tests/index-remove.bats

@@ -0,0 +1,44 @@
+#!/usr/bin/env bats
+# tests/index-remove.bats — index-append.py --remove mode.
+setup() {
+  load 'helpers'
+  export GENOMES_ROOT="${BATS_TEST_TMPDIR}"
+  g_src="$(make_fixture_genome)"; export g="$g_src"
+}
+
+@test "index --remove: deletes the matching entry, keeps the others" {
+  cd "$g"
+  python3 "$SKILL_SCRIPTS/index-append.py" --section Sources --entry '- [[sources/a]] — A. `maturity: draft`'
+  python3 "$SKILL_SCRIPTS/index-append.py" --section Sources --entry '- [[sources/b]] — B. `maturity: draft`'
+  grep -q 'sources/a' wiki/index.md
+  grep -q 'sources/b' wiki/index.md
+
+  run python3 "$SKILL_SCRIPTS/index-append.py" --remove 'sources/a'
+  [ "$status" -eq 0 ]
+  ! grep -q '\[\[sources/a\]\]' wiki/index.md
+  grep -q 'sources/b' wiki/index.md
+}
+
+@test "index --remove: idempotent when the entry is absent" {
+  cd "$g"
+  run python3 "$SKILL_SCRIPTS/index-append.py" --remove 'sources/does-not-exist'
+  [ "$status" -eq 0 ]
+  [[ "$output" == *'nothing to remove'* ]]
+}
+
+@test "index --remove: bumps last_updated" {
+  cd "$g"
+  python3 "$SKILL_SCRIPTS/index-append.py" --section Sources --entry '- [[sources/a]] — A. `maturity: draft`'
+  # set last_updated to an old date, then remove and check it moved
+  sed -i 's/^last_updated:.*/last_updated: 2000-01-01/' wiki/index.md
+  run python3 "$SKILL_SCRIPTS/index-append.py" --remove 'sources/a'
+  [ "$status" -eq 0 ]
+  ! grep -q '2000-01-01' wiki/index.md
+  grep -q "last_updated: $(date +%F)" wiki/index.md
+}
+
+@test "index --remove: rejects passing both --entry and --remove" {
+  cd "$g"
+  run python3 "$SKILL_SCRIPTS/index-append.py" --section Sources --entry '- [[sources/a]] — x' --remove 'sources/a'
+  [ "$status" -eq 2 ]
+}

tests/ingest-semantic.bats

@@ -0,0 +1,29 @@
+#!/usr/bin/env bats
+
+setup() {
+  load 'helpers'
+  source "$LIB_DIR/output.sh"
+  source "$LIB_DIR/lint.sh"
+}
+
+@test "lint tolerates source_path/source_sha256 in source frontmatter" {
+  G="$(make_fixture_genome)"
+  mkdir -p "$G/wiki/sources"
+  cat > "$G/wiki/sources/test-source.md" <<'EOFMD'
+---
+title: "Test Source"
+type: source
+domain: genome-test
+maturity: draft
+last_updated: 2026-06-25
+private: false
+tags: [test]
+source_path: raw/articles/test.md
+source_sha256: e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855
+---
+# Test Source
+body
+EOFMD
+  run lint_markdown_file "$G/wiki/sources/test-source.md" genome-test
+  [ "$status" -eq 0 ]
+}

tests/lint.bats

@@ -69,3 +69,80 @@ EOF
   run bash "$SKILL_SCRIPTS/scoped-lint.sh" genome-test wiki/sources/good.md
   [ "$status" -eq 0 ]
 }
+
+# --- duplicate-slug advisory (check_duplicates + its distance helpers) --------
+# These guard the dedup feature: correct edit-distance math, the warn-only
+# contract, the exact-self-match skip (run-ingest appends new slugs to the
+# index before lint runs), and that the advisory fires once per run, not once
+# per file.
+
+@test "levenshtein: identical strings have distance 0" {
+  run levenshtein cat cat
+  [ "$status" -eq 0 ]
+  [ "$output" -eq 0 ]
+}
+
+@test "levenshtein: kitten→sitting is 3 (textbook case)" {
+  run levenshtein kitten sitting
+  [ "$output" -eq 3 ]
+}
+
+@test "similarity: identical strings score 100" {
+  run similarity gpu-pricing gpu-pricing
+  [ "$output" -eq 100 ]
+}
+
+@test "check_duplicates: warns on a near-duplicate of an indexed concept" {
+  G="$(make_fixture_genome)"; cd "$G"
+  printf -- '- [[concepts/llm-routing]] — x\n' >> wiki/index.md
+  cat > .ingest-manifest.json <<'JSON'
+{"raw_source":"src","pages":[{"path":"wiki/concepts/llm-routings.md","status":"created"}]}
+JSON
+  run check_duplicates .ingest-manifest.json
+  [ "$status" -eq 0 ]
+  [[ "$output" == *"≈"* ]]
+  [[ "$output" == *"llm-routings"* ]]
+}
+
+@test "check_duplicates: silent when the new slug is unlike anything indexed" {
+  G="$(make_fixture_genome)"; cd "$G"
+  printf -- '- [[concepts/llm-routing]] — x\n' >> wiki/index.md
+  cat > .ingest-manifest.json <<'JSON'
+{"raw_source":"src","pages":[{"path":"wiki/concepts/budget-hardware.md","status":"created"}]}
+JSON
+  run check_duplicates .ingest-manifest.json
+  [ "$status" -eq 0 ]
+  [[ "$output" != *"≈"* ]]
+}
+
+@test "check_duplicates: an exact self-match is not flagged (index already has the slug)" {
+  G="$(make_fixture_genome)"; cd "$G"
+  # run-ingest step 1 inserts this run's slug into the index BEFORE lint runs;
+  # the slug must not be reported as a duplicate of itself.
+  printf -- '- [[concepts/llm-routing]] — x\n' >> wiki/index.md
+  cat > .ingest-manifest.json <<'JSON'
+{"raw_source":"src","pages":[{"path":"wiki/concepts/llm-routing.md","status":"created"}]}
+JSON
+  run check_duplicates .ingest-manifest.json
+  [ "$status" -eq 0 ]
+  [[ "$output" != *"≈"* ]]
+}
+
+@test "scoped-lint: duplicate advisory fires once across multiple files, not per file" {
+  G="$(make_fixture_genome)"
+  write_page "$G/wiki/concepts/data-pipelines.md" concept genome-test
+  write_page "$G/wiki/concepts/other-topic.md"    concept genome-test
+  printf -- '- [[concepts/data-pipeline]] — x\n' >> "$G/wiki/index.md"
+  cat > "$G/.ingest-manifest.json" <<'JSON'
+{"raw_source":"src","pages":[
+  {"path":"wiki/concepts/data-pipelines.md","status":"created"},
+  {"path":"wiki/concepts/other-topic.md","status":"created"}
+]}
+JSON
+  cd "$G"
+  export KG_LIB_DIR="$LIB_DIR"
+  run bash "$SKILL_SCRIPTS/scoped-lint.sh" genome-test \
+      wiki/concepts/data-pipelines.md wiki/concepts/other-topic.md
+  [ "$status" -eq 0 ]
+  [ "$(grep -c "≈" <<< "$output")" -eq 1 ]
+}

tests/open-pr-rolling.bats

@@ -0,0 +1,48 @@
+#!/usr/bin/env bats
+# open-pr-rolling.bats — a re-ingest of the same slug updates the OPEN PR's branch
+# (force-with-lease) instead of failing. Uses the local bare remote from make_fixture_genome.
+load helpers
+setup_file() { :; }
+
+@test "open-pr: re-ingest of the same slug rolls the branch forward (force-with-lease)" {
+  command -v jq >/dev/null 2>&1 || skip "jq not installed"
+  G="$(make_fixture_genome)"; cd "$G"
+  export FORGEJO_URL="http://forgejo.local" FORGEJO_USER=u FORGEJO_TOKEN=t DRY_RUN=1
+  body="$(mktemp)"; echo body > "$body"
+
+  # first ingest of slug x (v1)
+  mkdir -p wiki/sources; printf 'v1\n' > wiki/sources/x.md
+  run bash "$SKILL_SCRIPTS/open-pr.sh" --slug x --title "feat: ingest x" --body-file "$body" --base main
+  [ "$status" -eq 0 ]
+  git rev-parse --verify feat/ai-ingest-x
+  first="$(git rev-parse feat/ai-ingest-x)"
+
+  # simulate clean_start back to base, then an edited re-ingest (v2)
+  git switch -q main; git reset -q --hard origin/main; git clean -q -fd
+  printf 'v2-edited\n' > wiki/sources/x.md
+  run bash "$SKILL_SCRIPTS/open-pr.sh" --slug x --title "feat: ingest x" --body-file "$body" --base main
+  [ "$status" -eq 0 ]
+  second="$(git rev-parse feat/ai-ingest-x)"
+
+  # the branch was REBUILT from base (diverged), not appended: second is not a descendant of first
+  run git merge-base --is-ancestor "$first" "$second"
+  [ "$status" -ne 0 ]
+
+  # origin received the v2 content (force-with-lease pushed the rebuilt branch)
+  git fetch -q origin
+  run git show "origin/feat/ai-ingest-x:wiki/sources/x.md"
+  [ "$status" -eq 0 ]
+  [[ "$output" == *"v2-edited"* ]]
+}
+
+@test "open-pr: prune branch override still works after the rolling change" {
+  command -v jq >/dev/null 2>&1 || skip "jq not installed"
+  G="$(make_fixture_genome)"; cd "$G"
+  export FORGEJO_URL="http://forgejo.local" FORGEJO_USER=u FORGEJO_TOKEN=t DRY_RUN=1
+  body="$(mktemp)"; echo body > "$body"
+  mkdir -p wiki/sources; printf 'p\n' > wiki/sources/p.md
+  run bash "$SKILL_SCRIPTS/open-pr.sh" --branch "chore/prune-orphans-2026-06-30" \
+      --title "chore: prune 1 orphaned source(s)" --body-file "$body" --base main
+  [ "$status" -eq 0 ]
+  git rev-parse --verify "chore/prune-orphans-2026-06-30"
+}

tests/orphan-wiki.bats

@@ -0,0 +1,38 @@
+#!/usr/bin/env bats
+setup() {
+  load 'helpers'
+  export ORPHAN="${SKILL_SCRIPTS}/orphan-wiki.sh"
+  export GENOMES_ROOT="${BATS_TEST_TMPDIR}"
+  export INGEST_BASE="main"
+  export KG_LIB_DIR="${LIB_DIR}"            # orphan-wiki.sh sources clean-start.sh via KG_LIB_DIR
+  g_src="$(make_fixture_genome)"
+  export g_name="fixture-genome"
+  mv "$g_src" "${GENOMES_ROOT}/${g_name}"
+  export g="${GENOMES_ROOT}/${g_name}"
+  ( cd "$g" && rm -f raw/articles/test.md && git add -A && git commit -q -m "clear" && git push -q )
+}
+@test "orphan-wiki: no orphans when raw and source page match" {
+  mkdir -p "${g}/raw/articles"; echo "content" > "${g}/raw/articles/existing.md"
+  hash="$(sha256sum "${g}/raw/articles/existing.md" | cut -d' ' -f1)"
+  mkdir -p "${g}/wiki/sources"
+  printf -- '---\nsource_path: raw/articles/existing.md\nsource_sha256: %s\n---\n' "$hash" > "${g}/wiki/sources/existing.md"
+  ( cd "$g" && git add . && git commit -q -m "setup" && git push -q )
+  run bash "$ORPHAN" "$g_name"
+  [ "$status" -eq 0 ]; echo "$output" | jq -e '.count == 0'
+}
+@test "orphan-wiki: detects orphaned source page" {
+  mkdir -p "${g}/wiki/sources"
+  printf -- '---\nsource_path: raw/articles/deleted.md\nsource_sha256: abc123\n---\n' > "${g}/wiki/sources/orphaned.md"
+  ( cd "$g" && git add . && git commit -q -m "orphan" && git push -q )
+  run bash "$ORPHAN" "$g_name"
+  [ "$status" -eq 0 ]
+  echo "$output" | jq -e '.count == 1'
+  echo "$output" | jq -e '.detail[0].reason == "orphan"'
+}
+@test "orphan-wiki: ignores legacy pages without source_path" {
+  mkdir -p "${g}/wiki/sources"
+  printf -- '---\ntitle: "Legacy"\ntype: source\n---\n' > "${g}/wiki/sources/legacy.md"
+  ( cd "$g" && git add . && git commit -q -m "legacy" && git push -q )
+  run bash "$ORPHAN" "$g_name"
+  [ "$status" -eq 0 ]; echo "$output" | jq -e '.count == 0'
+}

tests/pending-raw.bats

@@ -0,0 +1,91 @@
+#!/usr/bin/env bats
+
+setup() {
+  load 'helpers'
+  export PENDING="${SKILL_SCRIPTS}/pending-raw.sh"
+  export GENOMES_ROOT="${BATS_TEST_TMPDIR}"
+  export INGEST_BASE="main"
+  export KG_LIB_DIR="${LIB_DIR}"
+
+  g_src="$(make_fixture_genome)"
+  export g_name="fixture-genome"
+  mv "$g_src" "${GENOMES_ROOT}/${g_name}"
+  export g="${GENOMES_ROOT}/${g_name}"
+
+  # FIX: make_fixture_genome ships raw/articles/test.md with no source page, which would
+  # otherwise count as a permanent 'new' and break every count assertion. Clear it so each
+  # test controls exactly what is pending (verified: count base becomes 0).
+  ( cd "$g" && rm -f raw/articles/test.md && git add -A \
+       && git commit -q -m "test: clear default raw" && git push -q )
+}
+
+@test "pending-raw: detects a brand new raw file" {
+  echo "new content" > "${g}/raw/articles/new-file.md"
+  ( cd "$g" && git add . && git commit -q -m "add raw" && git push -q )
+  run bash "$PENDING" "$g_name"
+  [ "$status" -eq 0 ]
+  echo "$output" | jq -e '.count == 1'
+  echo "$output" | jq -e '.detail[0].path == "raw/articles/new-file.md"'
+  echo "$output" | jq -e '.detail[0].reason == "new"'
+}
+
+@test "pending-raw: skips up-to-date files" {
+  echo "ok content" > "${g}/raw/articles/ok-file.md"
+  hash_ok="$(sha256sum "${g}/raw/articles/ok-file.md" | cut -d' ' -f1)"
+  cat > "${g}/wiki/sources/ok-file.md" <<FM
+---
+source_sha256: $hash_ok
+---
+FM
+  ( cd "$g" && git add . && git commit -q -m "add ok" && git push -q )
+  run bash "$PENDING" "$g_name"
+  [ "$status" -eq 0 ]
+  echo "$output" | jq -e '.count == 0'
+}
+
+@test "pending-raw: flags modified files" {
+  echo "content v1" > "${g}/raw/articles/mod-file.md"
+  hash_v1="$(sha256sum "${g}/raw/articles/mod-file.md" | cut -d' ' -f1)"
+  cat > "${g}/wiki/sources/mod-file.md" <<FM
+---
+source_sha256: $hash_v1
+---
+FM
+  ( cd "$g" && git add . && git commit -q -m "v1" && git push -q )
+  echo "content v2" > "${g}/raw/articles/mod-file.md"
+  ( cd "$g" && git add . && git commit -q -m "v2" && git push -q )
+  run bash "$PENDING" "$g_name"
+  [ "$status" -eq 0 ]
+  echo "$output" | jq -e '.count == 1'
+  echo "$output" | jq -e '.detail[0].reason == "modified"'
+}
+
+@test "pending-raw: nested subdirectory yields prefixed slug" {
+  mkdir -p "${g}/raw/articles/sub-b"
+  echo "subdir content" > "${g}/raw/articles/sub-b/file.md"
+  ( cd "$g" && git add . && git commit -q -m "subdir" && git push -q )
+  run bash "$PENDING" "$g_name"
+  [ "$status" -eq 0 ]
+  echo "$output" | jq -e '.count == 1'
+  echo "$output" | jq -e '.files[0] == "raw/articles/sub-b/file.md"'
+}
+
+@test "pending-raw: excludes noise (.stfolder, .gitkeep)" {
+  touch "${g}/raw/articles/.gitkeep"
+  mkdir -p "${g}/raw/articles/.stfolder"
+  touch "${g}/raw/articles/.stfolder/sync.log"
+  ( cd "$g" && git add . && git commit -q -m "noise" && git push -q )
+  run bash "$PENDING" "$g_name"
+  [ "$status" -eq 0 ]
+  echo "$output" | jq -e '.count == 0'
+}
+
+@test "pending-raw: reports both files on a slug collision" {
+  mkdir -p "${g}/raw/articles/cibo"
+  echo "c1" > "${g}/raw/articles/cibo-pane.md"
+  echo "c2" > "${g}/raw/articles/cibo/pane.md"
+  ( cd "$g" && git add . && git commit -q -m "collision" && git push -q )
+  run bash "$PENDING" "$g_name"
+  [ "$status" -eq 0 ]
+  echo "$output" | jq -e '.count == 2'
+}

tests/permissions.bats

@@ -0,0 +1,68 @@
+#!/usr/bin/env bats
+# tests/permissions.bats
+# Blinda i permessi del repo, cosi' un `cp`/deploy preserva l'eseguibilita' e non
+# ricapita il "Permission denied" (es. ingest-semantic.py lanciato diretto).
+#
+# Principio:
+#   - script con shebang lanciati direttamente  -> eseguibili (git mode 100755)
+#   - librerie *sourced* (lib/, providers/, registry.sh, globals.env) -> NON eseguibili (100644)
+
+REPO="${BATS_TEST_DIRNAME}/.."
+
+# Entry-point / script eseguibili (tutti hanno shebang; alcuni anche lanciati a mano per debug)
+EXECUTABLES=(
+  skills/ingest/scripts/ingest-semantic.py
+  skills/ingest/scripts/run-ingest.sh
+  skills/ingest/scripts/scoped-lint.sh
+  skills/ingest/scripts/open-pr.sh
+  skills/ingest/scripts/log-append.sh
+  skills/ingest/scripts/slug.sh
+  skills/ingest/scripts/pending-raw.sh
+  skills/ingest/scripts/orphan-wiki.sh
+  skills/ingest/scripts/index-append.py
+  scripts/add-genome.sh
+  scripts/setup.sh
+  scripts/setup-genomes.sh
+  scripts/setup-master.sh
+  scripts/lint-genomes.sh
+  scripts/verify-genomes.sh
+)
+
+# Librerie sourced: NON devono essere eseguibili.
+LIBRARIES=(
+  lib/lint.sh lib/output.sh lib/deps.sh lib/git-crypt.sh lib/scaffold.sh lib/structure.sh lib/clean-start.sh
+  providers/forgejo.sh providers/github.sh
+  registry.sh globals.env
+)
+
+git_mode() { git -C "$REPO" ls-files -s -- "$1" | awk '{print $1}'; }
+
+@test "executable scripts have the +x bit on disk" {
+  for f in "${EXECUTABLES[@]}"; do
+    [ -x "${REPO}/${f}" ] || { echo "NON eseguibile su disco: $f"; return 1; }
+  done
+}
+
+@test "executable scripts are recorded 100755 in git" {
+  for f in "${EXECUTABLES[@]}"; do
+    mode="$(git_mode "$f")"
+    [ -n "$mode" ]          || { echo "non tracciato in git: $f"; return 1; }
+    [ "$mode" = "100755" ]  || { echo "git mode $mode (atteso 100755): $f"; return 1; }
+  done
+}
+
+@test "sourced libraries are NOT executable in git (100644)" {
+  for f in "${LIBRARIES[@]}"; do
+    mode="$(git_mode "$f")"
+    [ -z "$mode" ] && continue          # non tracciato/opzionale -> salta
+    [ "$mode" = "100644" ] || { echo "git mode $mode (atteso 100644, e' sourced): $f"; return 1; }
+  done
+}
+
+@test "executable shell scripts pass bash -n (syntax)" {
+  for f in "${EXECUTABLES[@]}"; do
+    case "$f" in
+      *.sh) bash -n "${REPO}/${f}" || { echo "syntax error: $f"; return 1; } ;;
+    esac
+  done
+}

tests/raw-commit-quiet.bats

@@ -0,0 +1,75 @@
+#!/usr/bin/env bats
+# raw-commit-quiet.bats — quiet-window behaviour of genome-raw-commit.sh.
+# No Syncthing (no API key -> default author); pushes to a local bare repo via GENOME_PUSH_URL.
+setup() {
+  SCRIPT="${BATS_TEST_DIRNAME}/../deploy/nexus/genome-raw-commit.sh"
+  export HOME="${BATS_TEST_TMPDIR}/home"; mkdir -p "$HOME/.config"
+  root="${BATS_TEST_TMPDIR}/vaults"; mkdir -p "$root"
+  bare="${BATS_TEST_TMPDIR}/origin.git"; git init -q --bare "$bare"
+  cat > "$HOME/.config/knowledge-genome.env" <<EOF
+GENOME_VAULTS_ROOT=$root
+GENOME_BASE=main
+FORGEJO_USER=n8n-bot
+FORGEJO_HOST=127.0.0.1:3001
+FORGEJO_OWNER=Keru
+COMMITTER_NAME=n8n-bot
+COMMITTER_EMAIL=n8n-bot@homelab
+DEFAULT_AUTHOR_NAME=Tester
+DEFAULT_AUTHOR_EMAIL=tester@local
+EOF
+  export g="genome-test"; export vault="$root/$g"
+  git clone -q "$bare" "$vault" 2>/dev/null || mkdir -p "$vault"
+  ( cd "$vault"
+    git init -q 2>/dev/null || true
+    git config user.name n8n-bot; git config user.email n8n-bot@homelab; git config commit.gpgsign false
+    git checkout -q -b main 2>/dev/null || git switch -q main
+    mkdir -p raw/articles; echo seed > raw/articles/.gitkeep
+    git add -A; git commit -q -m init
+    git remote add origin "$bare" 2>/dev/null || git remote set-url origin "$bare"
+    git push -q -u origin main )
+  export GENOME_PUSH_URL="$bare"      # test seam -> push to the local bare repo
+}
+files() { ( cd "$vault" && git ls-files raw/ ) > "${BATS_TEST_TMPDIR}/f.txt"; }
+
+@test "raw-commit: holds a freshly-written raw, commits it once it settles" {
+  command -v jq >/dev/null 2>&1 || skip "jq not installed"
+  echo "still typing" > "$vault/raw/articles/hot.md"                       # fresh -> hot
+  echo "finished"     > "$vault/raw/articles/stable.md"
+  touch -d "10 minutes ago" "$vault/raw/articles/stable.md"                # settled
+
+  run bash "$SCRIPT" "$g"
+  [ "$status" -eq 0 ]
+  echo "$output" | jq -e '.status=="ok"'
+  files
+  grep -q 'raw/articles/stable.md' "${BATS_TEST_TMPDIR}/f.txt"             # committed
+  ! grep -q 'raw/articles/hot.md'  "${BATS_TEST_TMPDIR}/f.txt"            # held back
+
+  touch -d "10 minutes ago" "$vault/raw/articles/hot.md"                   # now it settles
+  run bash "$SCRIPT" "$g"
+  [ "$status" -eq 0 ]
+  files
+  grep -q 'raw/articles/hot.md' "${BATS_TEST_TMPDIR}/f.txt"               # now committed
+}
+
+@test "raw-commit: noop with held count while everything is still settling" {
+  command -v jq >/dev/null 2>&1 || skip "jq not installed"
+  echo "typing" > "$vault/raw/articles/wip.md"                            # fresh -> hot
+  run bash "$SCRIPT" "$g"
+  [ "$status" -eq 0 ]
+  echo "$output" | jq -e '.status=="noop"'
+  echo "$output" | jq -e '.held==1'
+}
+
+@test "raw-commit: a deletion is committed immediately (not subject to the quiet window)" {
+  command -v jq >/dev/null 2>&1 || skip "jq not installed"
+  # commit a settled file first
+  echo done > "$vault/raw/articles/old.md"; touch -d "10 minutes ago" "$vault/raw/articles/old.md"
+  run bash "$SCRIPT" "$g"; [ "$status" -eq 0 ]
+  files; grep -q 'raw/articles/old.md' "${BATS_TEST_TMPDIR}/f.txt"
+  # now delete it -> should commit the removal even though "just changed"
+  rm "$vault/raw/articles/old.md"
+  run bash "$SCRIPT" "$g"
+  [ "$status" -eq 0 ]
+  echo "$output" | jq -e '.status=="ok"'
+  files; ! grep -q 'raw/articles/old.md' "${BATS_TEST_TMPDIR}/f.txt"
+}

tests/run-ingest.bats

@@ -171,3 +171,41 @@ EOF
   [ "$status" -eq 0 ]
   [[ "$output" == *"develop"* ]]
 }
+
+@test "run-ingest: branch name matches slug.sh --raw for nested raw paths" {
+  command -v jq >/dev/null 2>&1 || skip "jq not installed"
+  G="$(make_fixture_genome)"; cd "$G"
+  mkdir -p wiki/sources
+  cat > wiki/sources/cibo-il-pane.md <<'EOFMD'
+---
+title: "Il Pane"
+type: source
+domain: genome-test
+tags: [cibo]
+maturity: draft
+last_updated: 2026-06-25
+private: false
+source_path: raw/articles/cibo/il-pane.md
+source_sha256: e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855
+---
+# Il Pane
+body
+EOFMD
+  cat > .ingest-manifest.json <<'EOFJSON'
+{
+  "raw_source": "raw/articles/cibo/il-pane.md",
+  "model": "qwen3.5-9b",
+  "reasoning": "Ingest.",
+  "pr_summary": "Ingest summary.",
+  "contradictions": "None",
+  "pages": [
+    {"path": "wiki/sources/cibo-il-pane.md", "summary": "Summary.", "maturity": "draft", "status": "created"}
+  ]
+}
+EOFJSON
+  export KG_LIB_DIR="$LIB_DIR"
+  export FORGEJO_URL="http://forgejo.local" FORGEJO_USER="u" FORGEJO_TOKEN="t" DRY_RUN=1
+  run bash "$SKILL_SCRIPTS/run-ingest.sh" genome-test
+  [ "$status" -eq 0 ]
+  [[ "$output" == *"cibo-il-pane"* ]]
+}

tests/run-prune.bats

@@ -0,0 +1,68 @@
+#!/usr/bin/env bats
+# tests/run-prune.bats — prune orphaned sources (no LLM, no network; DRY_RUN).
+setup() {
+  load 'helpers'
+  export PRUNE="${SKILL_SCRIPTS}/run-prune.sh"
+  export GENOMES_ROOT="${BATS_TEST_TMPDIR}"
+  export INGEST_BASE="main"
+  export KG_LIB_DIR="${LIB_DIR}"
+  export FORGEJO_URL="http://forgejo.local" FORGEJO_USER="u" FORGEJO_TOKEN="t"
+  export DRY_RUN=1
+  g_src="$(make_fixture_genome)"; export g_name="fixture-genome"
+  mv "$g_src" "${GENOMES_ROOT}/${g_name}"; export g="${GENOMES_ROOT}/${g_name}"
+  ( cd "$g" && rm -f raw/articles/test.md && git add -A && git commit -q -m clear && git push -q )
+}
+
+@test "run-prune: removes only the orphaned source + its index entry, opens a dry PR" {
+  command -v jq >/dev/null 2>&1 || skip "jq not installed"
+  cd "$g"
+  # kept: raw exists. orphan: raw missing.
+  echo content > raw/articles/kept.md
+  h="$(sha256sum raw/articles/kept.md | cut -d' ' -f1)"
+  printf -- '---\nsource_path: raw/articles/kept.md\nsource_sha256: %s\n---\nbody\n' "$h" > wiki/sources/kept.md
+  printf -- '---\nsource_path: raw/articles/gone.md\nsource_sha256: abc\n---\nbody\n' > wiki/sources/orphan.md
+  python3 "$SKILL_SCRIPTS/index-append.py" --section Sources --entry '- [[sources/kept]] — kept. `maturity: draft`'
+  python3 "$SKILL_SCRIPTS/index-append.py" --section Sources --entry '- [[sources/orphan]] — orphan. `maturity: draft`'
+  git add -A && git commit -q -m setup && git push -q
+
+  run bash "$PRUNE" "$g_name"
+  [ "$status" -eq 0 ]
+  [[ "$output" == *'"status":"ok"'* ]]
+  [[ "$output" == *'"count":1'* ]]
+
+  # only the orphan page is gone
+  [ ! -f wiki/sources/orphan.md ]
+  [ -f wiki/sources/kept.md ]
+  # index reflects the removal
+  ! grep -q 'sources/orphan' wiki/index.md
+  grep -q 'sources/kept' wiki/index.md
+  # committed on a chore/ branch (NOT feat/ai-ingest-*)
+  git rev-parse --verify "chore/prune-orphans-$(date +%F)"
+}
+
+@test "run-prune: no orphans -> count 0 and no PR/branch" {
+  command -v jq >/dev/null 2>&1 || skip "jq not installed"
+  cd "$g"
+  echo content > raw/articles/kept.md
+  h="$(sha256sum raw/articles/kept.md | cut -d' ' -f1)"
+  printf -- '---\nsource_path: raw/articles/kept.md\nsource_sha256: %s\n---\nbody\n' "$h" > wiki/sources/kept.md
+  git add -A && git commit -q -m setup && git push -q
+
+  run bash "$PRUNE" "$g_name"
+  [ "$status" -eq 0 ]
+  [[ "$output" == *'"count":0'* ]]
+  run git rev-parse --verify "chore/prune-orphans-$(date +%F)"
+  [ "$status" -ne 0 ]
+}
+
+@test "run-prune: refuses when an orphan path would escape wiki/ (defense in depth)" {
+  command -v jq >/dev/null 2>&1 || skip "jq not installed"
+  cd "$g"
+  # legacy page without source_path is ignored; a page with a missing raw is the orphan.
+  printf -- '---\nsource_path: raw/articles/gone.md\nsource_sha256: abc\n---\nbody\n' > wiki/sources/orphan.md
+  git add -A && git commit -q -m setup && git push -q
+  run bash "$PRUNE" "$g_name"
+  [ "$status" -eq 0 ]
+  [[ "$output" == *'"count":1'* ]]
+  [ ! -f wiki/sources/orphan.md ]
+}

tests/scripts.bats

@@ -86,3 +86,17 @@ EOF
   python3 "$SKILL_SCRIPTS/index-append.py" --section Sources --entry '- [[sources/foo]] — s. `maturity: draft`'
   grep -q "^last_updated: $(date +%F)$" wiki/index.md
 }
+
+@test "log-append: dedup on stable run_id prevents duplicate entries" {
+  G="$(make_fixture_genome)"; cd "$G"
+  stable_id="test-stable-run-id-001"
+  run bash "$SKILL_SCRIPTS/log-append.sh" --run-id "$stable_id" --type INGEST --subject "test" --model "m" \
+    --context "[[raw/x]]" --output "[[sources/x]]" --reasoning "r"
+  [ "$status" -eq 0 ]
+  run bash "$SKILL_SCRIPTS/log-append.sh" --run-id "$stable_id" --type INGEST --subject "test" --model "m" \
+    --context "[[raw/x]]" --output "[[sources/x]]" --reasoning "r"
+  [ "$status" -eq 0 ]
+  [[ "$output" == *"already present"* ]]
+  count="$(grep -cF "run_id: \`${stable_id}\`" wiki/log.md || true)"
+  [ "$count" -eq 1 ]
+}

tests/slug.bats

@@ -0,0 +1,30 @@
+#!/usr/bin/env bats
+
+setup() {
+  load 'helpers'
+  SLUG="${SKILL_SCRIPTS}/slug.sh"
+}
+
+@test "slug --raw: flat file remains unchanged" {
+  run bash "$SLUG" --raw "raw/articles/il-pane.md"
+  [ "$status" -eq 0 ]
+  [ "$output" = "il-pane" ]
+}
+
+@test "slug --raw: nested file gets folder prefix" {
+  run bash "$SLUG" --raw "raw/articles/cibo/il-pane.md"
+  [ "$status" -eq 0 ]
+  [ "$output" = "cibo-il-pane" ]
+}
+
+@test "slug --raw: distinct subdirs avoid collision" {
+  s1="$(bash "$SLUG" --raw "raw/articles/cibo/pane.md")"
+  s2="$(bash "$SLUG" --raw "raw/articles/storia/pane.md")"
+  [ "$s1" != "$s2" ]
+}
+
+@test "slug --raw: Bash and Python-calling-bash agree (single implementation)" {
+  b="$(bash "$SLUG" --raw "raw/articles/cibo/il-pane.md")"
+  p="$(python3 -c "import subprocess;print(subprocess.check_output(['bash','$SLUG','--raw','raw/articles/cibo/il-pane.md'],text=True).strip())")"
+  [ "$b" = "$p" ]
+}