scan-meta-project
Skill0 starsChanged yesterdayMITView on GitHub
Inspect repository meta-knowledge integrity, verify 100% boundary coverage, detect architectural drift, and opportunistically backfill missing facts without re-initialization.
In this repository19 other files it gives its agents.
- AGENTS.mdAGENTS.md
- templates/AGENTS.mdAGENTS.md
- decommission-meta-projectSkill
- ground-meta-projectSkill
- init-meta-projectSkill
- manage-meta-projectSkill
- record-observationSkill
- scan-meta-projectSkill
- sync-architectureSkill
- tune-meta-projectSkill
- update-meta-projectSkill
- decommission-meta-projectSkill
- ground-meta-projectSkill
- init-meta-projectSkill
- manage-meta-projectSkill
- record-observationSkill
- sync-architectureSkill
- tune-meta-projectSkill
- update-meta-projectSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
17 rules. About 0 are common, 6 are distinctive, and 11 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
11
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 36
No canonical copy → schedule migration (move legacy ➔ scripts/).
- Very specific: line 39
If neither script exists, verify boundaries manually against boundaries.md glob patterns.
- Distinctive: line 46
Verify presence of root AGENTS.md.
- Very specific: line 47
Ensure Section 1 contains North Stars and invariants (Circuit Breaker, Default Deny).
- Very specific: line 48
Ensure Section 2, 3, 4, 5 are populated without placeholder tags ({{...}}).
- Very specific: line 50
Verify the 5 canonical slices in .agents/project_knowledge/:
- Distinctive: line 63
Run the automated Boundary Gate audit with the resolved copy:
- Very specific: line 67
If any unmapped files or directories are reported, treat them as Black Zone breaches.
- Very specific: line 69
Verify that the active objective is clear and completed items are marked with - [x].
- Distinctive: line 78
Database & Storage: Read primary model/repository files.
- Distinctive: line 79
External Clients & APIs: Read integrations (AI SDKs, HTTP clients).
- Very specific: line 80
Containerization & CI: Read Dockerfile, docker-compose*.yml, .github/workflows/.
- Distinctive: line 81
Entrypoints & UI: Read main routers, CLI entrypoints, and command handlers.
- Very specific: line 83
Do NOT duplicate custom code parsing logic.
- Very specific: line 87
Check existing items in weak_spots.md: if a previously recorded technical debt is resolved, mark it as resolved or remove it.
- Distinctive: line 88
If new fragile couplings or architectural bottlenecks were discovered during the Core Inspection, record them.
- Very specific: line 102
Update tasks.md with new subtasks or completed milestones.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Repository Meta-Knowledge Integrity Scanner (Scan Meta-Project)
Operational skill for conducting fast express scans or deep-dive integrity audits of an existing meta-project setup without destructive re-initialization.
The skill operates in two modes:
--shallow(Express Sanity Check): Verifies existence and non-emptiness of rootAGENTS.md, 5 knowledge slices, executes Boundary Gate (check_boundaries.py --audit-tree) for 100% file coverage, and checks task freshness.--deep(Core Deep-Dive & Drift Check): Runs all shallow checks, invokes the Core Inspection Checklist across key runtime modules, delegates component/data-flow synchronization tosync-architecture, audits technical debt inweak_spots.md, and runs local integrity tests.
When to invoke skill
- The user requests: "scan-meta-project", "scan meta project", "инспекция мета-знаний", "чекни покрытие", "audit meta-knowledge", "check boundaries coverage", "найди слепые зоны", "verify slices".
- Pre-release verification, PR reviews, or resuming development after extensive changes made outside the agentic session.
- Automatically invoked by
init-meta-projectat the conclusion of initialization.
Execution steps
Phase 1: Mode Determination & Target Resolution
- Identify invocation mode from user prompt or default to
--shallow:--shallow: Routine integrity check, boundary coverage validation, task freshness.--deep: Full architectural drift detection, Core Inspection Checklist, test execution.
- Locate the Boundary Gate script deterministically:
- Canonical location first:
scripts/check_boundaries.py. - Legacy
.agents/scripts/check_boundaries.pyis supported ONLY as an explicit migration candidate: if found, compare with the canonical copy.- No canonical copy → schedule migration (move legacy ➔
scripts/). - Both exist and are IDENTICAL → migrate (delete legacy) at next update.
- Both exist and DIVERGE → HALT and report both paths with hashes; never silently use the first found.
- No canonical copy → schedule migration (move legacy ➔
- If neither script exists, verify boundaries manually against
boundaries.mdglob patterns.
- Canonical location first:
Phase 2: Shallow Inspection (--shallow)
- Passport Integrity:
- Verify presence of root
AGENTS.md. - Ensure Section 1 contains North Stars and invariants (
Circuit Breaker,Default Deny). - Ensure Section 2, 3, 4, 5 are populated without placeholder tags (
{{...}}).
- Verify presence of root
- Knowledge Slices Audit:
- Verify the 5 canonical slices in
.agents/project_knowledge/:architecture.md(components and data flows).boundaries.md(responsibility zones).weak_spots.md(fragile modules and risks).observations.md(5-column observation table).tasks.md(active objective and checklists).
- Verify the 5 canonical slices in
- Boundary Coverage (Zero Unmapped Dirs Rule):
- Locate the Boundary Gate utility deterministically:
- Canonical location first:
scripts/check_boundaries.py. - Legacy
.agents/scripts/check_boundaries.pyis supported ONLY as an explicit migration candidate: if found, compare with the canonical copy.- No canonical copy → schedule migration (move legacy ➔
scripts/). - Both exist and are IDENTICAL → migrate (delete legacy) at next update.
- Both exist and DIVERGE → HALT and report both paths with hashes; never silently use the first found.
- No canonical copy → schedule migration (move legacy ➔
- Canonical location first:
- Run the automated Boundary Gate audit with the resolved copy:
python scripts/check_boundaries.py --audit-tree - If any unmapped files or directories are reported, treat them as Black Zone breaches. Flag them immediately for inclusion in
boundaries.md.
- Locate the Boundary Gate utility deterministically:
- Task Register Check (
tasks.md):- Verify that the active objective is clear and completed items are marked with
- [x].
- Verify that the active objective is clear and completed items are marked with
Phase 3: Deep Inspection (--deep)
When executed with --deep, the agent runs all Phase 2 checks, followed by:
- Core Inspection Checklist (Key Modules):
- Database & Storage: Read primary model/repository files. Inspect schema coupling, migrations, and monolithic growth.
- External Clients & APIs: Read integrations (AI SDKs, HTTP clients). Check timeout settings, circuit breakers, and error propagation.
- Containerization & CI: Read
Dockerfile,docker-compose*.yml,.github/workflows/. Verify build matrix and security flags. - Entrypoints & UI: Read main routers, CLI entrypoints, and command handlers. Check platform payload constraints and input validation.
- Architectural Drift Detection (Delegation to
sync-architecture):- Do NOT duplicate custom code parsing logic. Invoke the procedural skill
sync-architecture:sync-architectureperforms Pre-Flight Preservation of alllocal_repo_datablocks.- Updates
.agents/project_knowledge/architecture.mdwith current components and End-to-End Data Flows.
- Do NOT duplicate custom code parsing logic. Invoke the procedural skill
- Risk & Debt Recalibration (
weak_spots.md):- Check existing items in
weak_spots.md: if a previously recorded technical debt is resolved, mark it as resolved or remove it. - If new fragile couplings or architectural bottlenecks were discovered during the Core Inspection, record them.
- Check existing items in
- Automated Test Run:
- Execute project test suite using standard test discovery:
python -m unittest discover -s tests -p "test_*.py" -v # Or stack-specific runner (e.g. pytest, npm test, cargo test)
- Execute project test suite using standard test discovery:
Phase 4: Opportunistic Refresh & Report
- Backfill Identified Gaps (Read + Opportunistic Refresh):
- Atomically write discovered unmapped paths into
boundaries.md(within<!-- local_repo_data:boundaries -->). - Update
tasks.mdwith new subtasks or completed milestones. - Record newly resolved environment quirks into
observations.md.
- Atomically write discovered unmapped paths into
- Deliver Scan Report:
Present a concise summary to the user:
### 🔍 Scan Report (Meta-Project Scanner) - **Mode:** `--shallow` | `--deep` - **Passport & Slices:** 5/5 slices present, AGENTS.md valid - **Boundary Coverage:** 100% mapped (or N unmapped paths detected) - **Architectural Drift:** In sync (or updated via sync-architecture) - **Identified Action Items:** [List of 1–3 concrete recommendations, if any]
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

