How this command is triggered — by the user, by Claude, or both
Slash command
/eng:reflectThe summary Claude sees in its command listing — used to decide when to auto-load this command
# Self-Improvement and Reflection Review the current session and identify learnings that should be captured for future sessions. ## Instructions 1. **Review the session**: Look back at the conversation and identify: - Mistakes made and how they were resolved - Patterns that worked well - Knowledge gaps that caused inefficiency - User corrections or feedback 2. **Categorize learnings**: - **Critical rules**: Things that must always/never be done - **Patterns**: Reusable approaches for common tasks - **Domain knowledge**: Codebase-specific information - **Debugging...
Review the current session and identify learnings that should be captured for future sessions.
Review the session: Look back at the conversation and identify:
Categorize learnings:
Update documentation:
CLAUDE.md under appropriate sectionsCLAUDE.md files in subdirectoriesConsider scope - where should the improvement go?
Repo-specific (CLAUDE.md in the repo):
Directory-specific (CLAUDE.md in subdirectory):
Standalone/CLAUDE.md for standalone test patterns)AI Skills repo (PostSharp.Engineering.AISkills):
Format:
Summarize:
npx claudepluginhub postsharp-ops/postsharp.engineering.aiskills --plugin eng/compoundReflects on the current session to extract tacit knowledge and friction points, producing a summary of learnings or confirming none exist.
/learnMines the current conversation for recurring corrections and durable lessons, then proposes additions to CLAUDE.md to prevent repeating mistakes.
/learnRecords lessons learned from errors, sessions, or manual input, and suggests automation opportunities by analyzing git history. Also supports listing, editing, and removing entries.
/reflectGuides reflection on recent work to extract and record learnings, root causes, and decisions using git log and task tracking commands.
/learnExtracts reusable knowledge and patterns from the current Claude Code session, evaluates quality, and updates project documentation and rules. Supports --status, --eval, --export, and --import subcommands.
/review-gladiatorReflects on accumulated observations and session history to cluster patterns and generate actionable recommendations for workflow improvement. Optionally filters reflection to a specific topic.