Stats
Actions
Tags
Help us improve
Share bugs, ideas, or general feedback.
From context-files
Review the current session for friction, repeated mistakes, and actionable improvements. Classifies findings as fix-in-codebase, add-tooling, add-to-context-file, or one-off. Run at the end of or during a session.
npx claudepluginhub maxnoller/claude-code-plugins --plugin context-filesHow this skill is triggered — by the user, by Claude, or both
Slash command
/context-files:session-reviewThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Session ID: ${CLAUDE_SESSION_ID}
Analyzes current Claude Code session for agent efficiency (tool precision, autonomy) and quality (CLAUDE.md compliance, code patterns), scoring dimensions and surfacing 2-3 actionable improvements.
Analyzes AI coding session transcripts from Claude or Codex to generate structured insights on prompt quality, strategy critique, key decisions, and takeaways for improvement.
Reviews Claude Code session transcripts for prompting effectiveness, agent performance, and environment gaps, producing actionable recommendations. Invoke via /review-session for current or specified path.
Share bugs, ideas, or general feedback.
Session ID: ${CLAUDE_SESSION_ID}
Find the session file:
~/.claude/projects/**/${CLAUDE_SESSION_ID}.jsonl
Use Grep to extract conversation turns (lines containing "type":"user" or "type":"assistant") and read them to reconstruct what happened in the session.
Look for:
Repeated corrections — things the agent got wrong more than once or needed multiple attempts. Each is a signal that something is unclear in the codebase, tooling, or context files.
Wrong assumptions — cases where the agent inferred something incorrectly. Could the code itself be clearer so the right answer is obvious?
Workflow friction — steps that were manual, slow, or awkward. Could a script, hook, or MCP tool automate this?
Missing affordances — things the agent had to ask about or look up that should have been self-evident. Missing types, unclear names, undocumented edge cases, missing .env.example.
Context file candidates — only flag for CLAUDE.md / AGENTS.md if it's a consistent mistake that cannot be fixed in the codebase. This should be rare.
What worked well — patterns worth preserving.
For each finding, assign exactly one of:
Output format:
Findings (prioritized by impact, most actionable first):
What worked well (brief)
Prefer fixing root causes over patching instructions. Every "Fix in codebase" finding is strictly better than "Add to context file". A well-typed codebase with clear naming needs almost no CLAUDE.md.