05-documentation
Skill1 starsChanged 53 days agoMITView on GitHub
Version: 1.0.0 Compatible: project_skills >= 1.0 Requires: discovery Outputs: documentation-map.md, and any documents from templates/ as appropriate Establish and maintain living documentation for the project. Every agent must leave the project better documented than it found it. Documentation is not a one-time task. It is a continuous obligation. Do not create every document. Create the minimum set that allows a person (or agent) to understand, operate, maintain, and safely change the project. When operating under Tier 4 Enterprise Controlled standard,…
We split this skill into the rules it gives an agent
15 rules. About 0 are common, 11 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈11
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 20
Do not create every document.
- Very specific: line 33
Strict Verification Lifecycle: Every feature, fix, or plan item must transition through: DISCOVERED $\to$ PLANNED $\to$ IN_PROGRESS $\to$ …
- Distinctive: line 41
When operating under Tier 4 Enterprise Controlled standard, maintain these exact synchronized files:
- Distinctive: line 68
Every project using Agent Blueprint should follow this documentation structure:
- Distinctive: line 120
Not every project needs every directory.
- Distinctive: line 127
Every document has a last-reviewed date and reviewer.
- Distinctive: line 128
Every document that describes behavior must be updateable by the agent that changes the behavior.
- Distinctive: line 129
If code and documentation disagree, the agent must resolve the disagreement — not assume the documentation is wrong.
- Distinctive: line 131
Do not duplicate information across documents.
- Very specific: line 140
Before modifying any code, the agent MUST create or update TASK.md (or active task tracker) detailing:
- Distinctive: line 143
Never make blind edits without a visible plan.
- Distinctive: line 146
If the task introduces or alters system behavior, the agent MUST update the corresponding living document before closing the task:
- Very specific: line 153
Every agent MUST append an entry to CHANGELOG.md (following the Keep a Changelog standard):
- Very specific: line 170
Run all automated unit and integration tests (cargo test, npm test, pytest, etc.).
- Distinctive: line 171
Ensure zero compile errors, zero linter regressions, and clean test runs before handing back control.
≈ “Common” is an estimate: at least 1 in 5 of the 308,232 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
05-Documentation Skill
Version: 1.0.0 Compatible: project_skills >= 1.0 Requires: discovery Outputs: documentation-map.md, and any documents from templates/ as appropriate
Purpose
Establish and maintain living documentation for the project. Every agent must leave the project better documented than it found it.
Documentation is not a one-time task. It is a continuous obligation.
Principle: Minimum Sufficient Documentation
Do not create every document. Create the minimum set that allows a person (or agent) to understand, operate, maintain, and safely change the project.
Documentation Tiers
| Tier | Project Type | Documents |
| 1 — Minimal | Script, small tool, simple service | README.md |
| 2 — Standard | Small web app, API, internal tool | README + PROJECT + ARCHITECTURE + DATABASE + API + SECURITY |
| 3 — Full | SaaS, marketplace, commercial product | Tier 2 + BUSINESS + BUSINESS-MODEL + ACTORS + TESTING + DEPLOYMENT + OPERATIONS |
| 4 — Enterprise Controlled | High-scale platforms, SaaS, mission-critical systems | The 19 Controlled Documents (00_MASTER_INDEX to 18_FEATURE_CHECKLIST) |
Cardinal Operating Principles:
- Evidence Over Documentation: Code and runtime evidence are stronger than documentation claims. If documentation claims a feature is complete but implementation or tests do not prove it, mark it as
INCOMPLETEorPENDING. - Strict Verification Lifecycle: Every feature, fix, or plan item must transition through:
DISCOVERED$\to$PLANNED$\to$IN_PROGRESS$\to$RUNTIME_EVIDENCE$\to$VERIFIED_DONE. - Continuous Synchronization: Documentation must never drift from actual code.
The 19 Controlled Documents Suite (00–18)
When operating under Tier 4 Enterprise Controlled standard, maintain these exact synchronized files:
README.md— System identity, badges, quickstart, and environment setup00_MASTER_INDEX.md— Master documentation registry and navigation hub01_PROJECT_BASELINE.md— Vision, scope boundaries, problem statement, and goals02_ARCHITECTURE.md— System topology, modular monolith structure, data flows03_PORTALS.md— Portal workflows (Public Directory, Website Studio, Owner Dashboard, Admin Launcher)04_NAVIGATION_MASTER.md— Site tree, Next.js routing, and navigation architecture05_LEGACY_FEATURE_MAPPING.md— Legacy code audit, technical debt, and decommission schedule06_MODULE_CATALOG.md— Engine modules, service layers, and crate boundaries07_DATA_MODEL_MASTER.md— PostgreSQL schemas, PostGIS models, ERD, and migrations (001–018)08_FINANCIAL_CORE.md— Plans catalog, Razorpay billing, recurring subscriptions, invoices09_SECURITY_RBAC_WORKFLOW.md— Argon2id, CSRF protection, sliding rate limiter, SSRF SafeFetcher10_API_AND_INTEGRATIONS.md— REST API reference, webhooks, and third-party adapters (OSM, Google Places)11_JOBS_AND_AUTOMATION.md— Background daemons (Scanner Cron, Notification Worker)12_TESTING_AND_VERIFICATION.md— Automated test suite matrix and verification commands13_IMPLEMENTATION_ROADMAP.md— Phase progression from Phase 0 to Phase 514_PROGRESS_TRACKER.md— Real-time progress tracker with checkbox status15_GAP_AND_RISK_REGISTER.md— Known limitations, risk mitigations, and technical debt16_DEFINITION_OF_DONE.md— Non-negotiable quality and release criteria17_CHANGELOG.md— Release history and session audit trail18_FEATURE_CHECKLIST.md— Comprehensive feature matrix with code-level proof of implementation
Documentation Standard
Every project using Agent Blueprint should follow this documentation structure:
docs/
├── 00-project/
│ ├── overview.md # What this project is
│ ├── goals.md # Why it exists, what it aims to achieve
│ ├── scope.md # What is in scope and out of scope
│ └── glossary.md # Domain terms
│
├── 01-business/
│ ├── business-model.md
│ ├── actors.md
│ ├── processes.md
│ ├── business-rules.md
│ └── state-machines.md
│
├── 02-architecture/
│ ├── architecture.md
│ ├── domains.md
│ ├── data-flow.md
│ ├── integrations.md
│ └── decisions/ # ADRs
│
├── 03-engineering/
│ ├── code-structure.md
│ ├── database.md
│ ├── api.md
│ └── frontend.md
│
├── 04-security/
│ └── security.md
│
├── 05-testing/
│ └── testing.md
│
├── 06-performance/
│ └── performance.md
│
├── 07-deployment/
│ └── deployment.md
│
├── 08-operations/
│ └── operations.md
│
├── 09-audits/
│ └── [audit reports]
│
└── 10-roadmap/
└── roadmap.mdNot every project needs every directory. Create only what applies.
Documentation Rules
- Documentation lives in
docs/or at the project root for top-level files (README, PROJECT.md, etc.). - Every document has a last-reviewed date and reviewer.
- Every document that describes behavior must be updateable by the agent that changes the behavior.
- If code and documentation disagree, the agent must resolve the disagreement — not assume the documentation is wrong.
- Outdated documentation is worse than no documentation. When in doubt, update or remove.
- Do not duplicate information across documents. Reference instead.
Mandatory Agent Protocol: Task Tracker & Changelog Audit
Whenever an AI agent performs ANY change on a project, it MUST follow this strict 4-step protocol:
Step 1: Pre-Flight Task List (TASK.md)
Before modifying any code, the agent MUST create or update TASK.md (or active task tracker) detailing:
- Planned changes broken into discrete checklist items:
- [ ] Task item - Mark tasks
- [x]in real time as work is completed. - Never make blind edits without a visible plan.
Step 2: Living Documentation Sync (docs/)
If the task introduces or alters system behavior, the agent MUST update the corresponding living document before closing the task:
- New/modified endpoints $\to$ update
docs/03-engineering/api.md - New/modified database models or migrations $\to$ update
docs/03-engineering/database.md - Changes to pricing or business rules $\to$ update
docs/01-business/business-model.md - Architecture or infrastructure shifts $\to$ create an ADR in
docs/02-architecture/decisions/
Step 3: Mandatory Changelog Audit Trail (CHANGELOG.md)
Every agent MUST append an entry to CHANGELOG.md (following the Keep a Changelog standard):
## [Unreleased] - YYYY-MM-DD
### Added
- [Feature description] (Files: `path/to/file`)
### Changed
- [Modification description] (Files: `path/to/file`)
### Fixed
- [Bug fix description] (Files: `path/to/file`)
### Documentation
- Updated `docs/...` to reflect [change]Step 4: Verification & Test Rigor
- Run all automated unit and integration tests (
cargo test,npm test,pytest, etc.). - Ensure zero compile errors, zero linter regressions, and clean test runs before handing back control.
What Every Agent Must Verify Before Exit
TASK.mdhas all completed items checked off[x]CHANGELOG.mdhas a clear audit trail entry of changes made- Living documentation in
docs/reflects the new reality - All automated tests pass with 0 failures
- No unneeded dependencies or files were created
Output
documentation-map.md
Documentation Map
=================
Tier: [1 | 2 | 3 | 4]
Documents:
- [path]: [status: present | missing | outdated], [last reviewed]
Gaps:
- [document that should exist but does not]
Outdated:
- [document that exists but is outdated]In this repository17 other files it gives its agents.
- AGENTS.mdAGENTS.md
- llms.txtllms.txt
- 00-orchestratorSkill
- 01-discoverySkill
- 02-project-contextSkill
- 03-business-architectureSkill
- 04-architectureSkill
- 06-codebase-auditSkill
- 07-securitySkill
- 08-testingSkill
- 09-performanceSkill
- 10-auditSkill
- 11-controlled-source-of-truthSkill
- 12-context-engineeringSkill
- 13-phase-loop-deliverySkill
- 14-forensics-and-debuggingSkill
- 15-autonomous-loop-and-simplificationSkill
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
Reviews can't be read right now.
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.

