From coding-agent-orchestration-harness
Self-improvement loop for Orchestrator-driven workflows. Use immediately after any user correction or behavior change request (workflow, validation, scope, tone, output format), after missed required gates (e.g., validation), or after review misses. Requires completing a post-correction micro-checklist before ending the turn.
How this skill is triggered — by the user, by Claude, or both
Slash command
/coding-agent-orchestration-harness:improvement-loopThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
This skill standardizes the “after correction → prevent recurrence” loop.
This skill standardizes the “after correction → prevent recurrence” loop.
It introduces a repo-local Lessons Log and a promotion mechanism:
docs/coding-agent/lessons.mddocs/coding-agent/rules/*.md)docs/coding-agent/lessons.mdIf prevention is unclear:
docs/coding-agent/lessons.md and apply them proactively.If you need the post-correction checklist:
If you need a lessons file template:
If you need a single lesson entry template:
If you need guidelines for promoting lessons to rules/skills:
npx claudepluginhub ebigunso/agent-harness --plugin coding-agent-orchestration-harnessCaptures friction from coding sessions like user corrections, redos, and frustration to log lessons learned and route improvements to context files. Triggers on /retrospective or detected signals.
Logs errors, user corrections, missing features, API failures, knowledge gaps, and best practices to .learnings/ markdown files. Promotes key insights to CLAUDE.md and AGENTS.md for AI agent self-improvement.
Analyzes skill outcome logs and user corrections, then proposes self-improvements to skill behavior and description hygiene.