ahtishamshahzad/agent_dev_flow/documentation
- community
Checking that the SKILL.md file is still in the repository.
Use to write or update documentation of the work inside `.ai/` — decisions, project state, work items, knowledge, generated reports — concisely and in its canonical place. For the applications' own documentation (the `docs/` tree), use `application-documentation` instead.
Install
npx vskill@latest install ahtishamshahzad/agent_dev_flow/documentationSKILL.md
Documentation
Purpose
Produce and maintain documentation of the work — placed canonically, kept concise, and never duplicated. Implements ../../system/DOCUMENTATION_RULES.md.
Scope: everything under .ai/. The applications' own documentation — the docs/ tree with a folder per app and a file per screen, page, endpoint, or job — belongs to application-documentation. Both are needed; neither restates the other.
When to Use
- Recording decisions, project state, work items, knowledge, or reports.
- Updating READMEs/indexes when structure changes.
- Not for documenting the applications themselves — screens, pages, endpoints, jobs go to
application-documentation. - Not to pad output with narration or restate canonical rules.
Inputs
- The fact/decision/state to document and its canonical home.
- Existing docs that may already cover it.
Discovery Questions
- Is this canonical (
../../system/), project state (../../projects/current/), knowledge (../../knowledge/), or generated (../../generated/)? - Or is it a fact about the built product? Then it is
docs/, not.ai/— hand toapplication-documentation. - Does a doc already cover this (update vs create)?
- Who reads this next, and what do they need?
Responsibilities
- Place docs in the correct canonical location (see the map in
DOCUMENTATION_RULES.md). - Keep adapters thin — link to
.ai/, never duplicate it. - Record decisions and their reasons (including rejected alternatives).
- Keep progress/task notes concise (status, decisions, links).
- Update relevant indexes (e.g.
../README.md) when adding entries. - Archive stale generated content (
../../generated/).
Required Workflow
- Identify the canonical home.
- Check for an existing doc to update.
- Write concisely; capture the "why."
- Update the relevant index.
- Archive superseded generated docs.
Decision Rules
- Single source of truth: document once, link elsewhere.
- Update over duplicate; delete/correct stale over contradict.
- Generated reports are archivable, not permanent context.
Rules
- No duplication of system rules into adapters or other docs.
- No secrets/unredacted PII in any doc (
../../system/SECURITY_RULES.md). - Write for reload: self-locating, with links.
Anti-Patterns
- Fat adapters restating the system.
- Verbose progress logs that bury the decision.
- Duplicate docs that drift out of sync.
- Long code examples in permanent skills (use
../../references//../../templates/).
Validation Checklist
- [ ] Correct canonical location.
- [ ] Concise; decisions + reasons recorded.
- [ ] No duplication of canonical content.
- [ ] Index updated.
- [ ] Stale generated docs archived.
- [ ] No secrets/PII.
Definition of Done
Documentation placed canonically, concise, non-duplicative, with decisions and reasons recorded, indexes updated, and stale generated content archived.
Related Skills
application-documentation (the docs/ half), project-orchestrator, code-review, final-quality-audit, and any skill producing records.
Related Knowledge
../../knowledge/ (where durable knowledge lands).
Related References
../../references/ structure conventions.
Context Loading Guidance
- Requires: the item to document and its canonical home.
- Does not require: unrelated source, the full reference tree, other skills' bodies.
- May load: the target index/README to update.
- Stop when: the doc + index are updated and stale content archived.
Token Efficiency Guidance
Write the minimum that serves the next reader. Link rather than paste. Summarize source into decisions.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryChecking.ai/skills/documentation/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/ahtishamshahzad/agent_dev_flow/documentation)