Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
End-of-session automation that runs in four phases: commits and pushes uncommitted work, extracts key decisions from the session, uses gpt-4o-mini to detect repeated patterns and automation opportunities, then writes learnings to dated memory files and updates project docs. Can be triggered manually or at session end.
Combines git hygiene, session reflection, and pattern detection into one command instead of requiring separate commit, note-taking, and review steps.
When to use it
- Committing forgotten staged files before closing a project
- Logging decisions made during a long debugging session
- Spotting a repeated question that signals a missing automation
- Updating AGENTS.md with a workflow fix discovered mid-session
- Reviewing what actually got done at the end of a work block
Example Workflow
Here's how your AI assistant might use this skill in practice.
User says: wrap up
- 1Check workspace for unstaged or uncommitted files and commit with an auto-generated message
- 2Push committed changes to the remote
- 3Scan session conversation and recent memory entries for key decisions and outcomes
- 4Send extracted learnings to gpt-4o-mini to identify repeated mistakes or automation opportunities
- 5Write a dated learnings file and update AGENTS.md or MEMORY.md if new patterns were found
Commit confirmation, a 1-3 bullet learnings summary, any detected patterns with suggested actions, and a list of updated files
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.