From awesome-kit
End-of-session documentation review - evaluates session work and recommends documentation updates
How this skill is triggered — by the user, by Claude, or both
Slash command
/awesome-kit:update-documentationThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Evaluate the current session's work and recommend documentation updates that will make future sessions more efficient. This skill translates session experience into persistent improvements across CLAUDE.md files, skills, and reference documents.
Evaluate the current session's work and recommend documentation updates that will make future sessions more efficient. This skill translates session experience into persistent improvements across CLAUDE.md files, skills, and reference documents.
Run at end of session before committing/pushing, when there is meaningful work to capture.
Review the session for three categories of documentable events:
Code changes - new or modified interfaces, scripts, libraries, configurations
git diff --stat and git log --oneline for this session's commitsRecovered errors - tool errors, incorrect actions, misunderstood requests
Inefficient operations - expensive exploration to find information
Apply these filters to each candidate. Drop items that fail:
| Filter | Keep if... |
|---|---|
| Recurrence | Likely to happen again in future sessions |
| Consequence | Cost more than 2-3 tool calls to resolve |
| Not already documented | Information isn't already in CLAUDE.md, skills, or docs |
| Not derivable | Can't be found in 1-2 obvious tool calls (e.g., git log, reading a file header) |
Place each surviving candidate into a documentation target:
| Target | When appropriate | Size guideline |
|---|---|---|
Root ~/.claude/CLAUDE.md | Universally needed for navigation or avoiding common mistakes | Keep under 200 lines total |
Directory CLAUDE.md | Specific to files in that directory/subdirectory | Keep under 60 lines |
Skill SKILL.md | Within an existing skill's domain | Keep under 200 lines |
| Skill reference doc | Sub-domain detail not always needed | No hard limit; loaded conditionally |
Classification rules:
For each candidate, check the current state of the target document:
Output a table of recommended changes:
| # | Target file | Change type | Summary | Severity |
|---|------------|-------------|---------|----------|
| 1 | ~/.claude/CLAUDE.md | Add line | Quick ref for new script X | Medium |
| 2 | ~/.claude/skills/foo/SKILL.md | Add section | Pattern for handling Y | High |
Severity levels:
Wait for user approval before making changes. User may accept all, select specific items, or modify recommendations.
IN SCOPE:
OUT OF SCOPE:
Provides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub kitaekatt/plugins-kit --plugin awesome-kit