How do I install the Maintain Docs skill?

Run npx skillmds add majiayu000/maintain-docs-2 in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Maintain Docs safe to use?

This skill has not completed SkillMD's automated safety review yet. Capability flags: docs only. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Maintain Docs?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Maintain Docs free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Maintain Docs?

majiayu000 (@majiayu000) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Maintain Docs by majiayu000 — Agent Skill

Name: Maintain Docs
Rating: 4.2 (5 reviews)
Author: majiayu000

File contents

Process Feedback and Maintain Documentation

Phase 1: Evaluate session summaries

Read all .md files in feedback/summaries/ that haven't been processed yet
- Check feedback/issue-tracker.md to see which summaries are already listed as sources
For each new summary, identify doc/skill issues: incorrect information, missing guidance, confusing instructions, stale examples
If deeper investigation is needed, read the raw transcript in feedback/transcripts/ (if still on disk)
Record findings in feedback/issue-tracker.md with file path, issue description, and source summary

Phase 2: Review skill usage

Read feedback/skill-usage.log if it exists
Identify:
- Skills never invoked (candidates for removal or better descriptions)
- Skills invoked frequently (candidates for optimization)
- Skills with consistent post-invocation errors in summaries (need fixing)
Record findings in feedback/issue-tracker.md

Phase 3: Apply fixes (frequency-gated)

Read feedback/issue-tracker.md
Only act on issues reported by 2+ independent sessions (or 1 session with concrete evidence like a build failure)
For each actionable issue:
- Read the referenced doc/skill file
- Only edit sections within  markers (or add new auto-managed sections)
- Make minimal, targeted fixes
If an issue requires substantial work, create a document in docs/ describing the problem and proposed fix instead

Phase 4: Prune and consolidate

Scan docs/skills for redundant or conflicting instructions — flag for human review
Print summary of all changes made and issues deferred

Constraints

Do NOT edit content outside  markers
Do NOT act on single-occurrence reports — require 2+ independent sessions (or 1 with concrete build/test failure evidence)
Do NOT delete documentation or skills — only flag candidates for human review
Do NOT rewrite entire files — make minimal, targeted fixes
Do NOT maintain .claude/HISTORY.md — that is the history-keeper agent's responsibility

---name: maintain-docs-2description: Process session summaries and skill usage logs, identify documentation issues, and apply frequency-gated fixes. Use periodically to maintain project documentation quality.--- # Process Feedback and Maintain Documentation ## Phase 1: Evaluate session summaries 1. Read all `.md` files in `feedback/summaries/` that haven't been processed yet - Check `feedback/issue-tracker.md` to see which summaries are already listed as sources 2. For each new summary, identify doc/skill issues: incorrect information, missing guidance, confusing instructions, stale examples 3. If deeper investigation is needed, read the raw transcript in `feedback/transcripts/` (if still on disk) 4. Record findings in `feedback/issue-tracker.md` with file path, issue description, and source summary ## Phase 2: Review skill usage 5. Read `feedback/skill-usage.log` if it exists 6. Identify: - Skills never invoked (candidates for removal or better descriptions) - Skills invoked frequently (candidates for optimization) - Skills with consistent post-invocation errors in summaries (need fixing) 7. Record findings in `feedback/issue-tracker.md` ## Phase 3: Apply fixes (frequency-gated) 8. Read `feedback/issue-tracker.md` 9. Only act on issues reported by **2+ independent sessions** (or 1 session with concrete evidence like a build failure) 10. For each actionable issue: - Read the referenced doc/skill file - Only edit sections within `` markers (or add new auto-managed sections) - Make minimal, targeted fixes 11. If an issue requires substantial work, create a document in `docs/` describing the problem and proposed fix instead ## Phase 4: Prune and consolidate 12. Scan docs/skills for redundant or conflicting instructions — flag for human review 13. Print summary of all changes made and issues deferred ## Constraints - Do NOT edit content outside `` markers - Do NOT act on single-occurrence reports — require 2+ independent sessions (or 1 with concrete build/test failure evidence) - Do NOT delete documentation or skills — only flag candidates for human review - Do NOT rewrite entire files — make minimal, targeted fixes - Do NOT maintain `.claude/HISTORY.md` — that is the history-keeper agent's responsibility

Frequently asked questions