Automated harness evolution for AI agents. A Claude Code plugin that iteratively optimizes system prompts, routing, retrieval, and orchestration code using full-trace counterfactual diagnosis. Based on Meta-Harness (Lee et al., 2026).
Install
npx skills add raphaelchristi/harness-evolvergh skill install raphaelchristi/harness-evolvergit clone --depth 1 https://github.com/raphaelchristi/harness-evolverSkills in this collection
- dev:dry-runRaphael ValdetaroUse when the user wants to smoke-test the evolve pipeline, test tools, or verify the plugin works end-to-end. Also use when the user says 'dry run', 'smoke test', or 'test pipeline'.
- dev:releaseRaphael ValdetaroUse when the user wants to release a new version, publish to npm, create a GitHub release, bump version, or tag a release. Also use when the user says 'release', 'publish', 'bump version', 'tag', or 'npm publish'.
- dev:validateRaphael ValdetaroUse when the user wants to validate the plugin, check integrity, verify cross-references, or before a release. Also use when the user says 'validate', 'check plugin', or 'verify'.
- harness:certifyRaphael ValdetaroUse when the user wants to verify that the evolved agent's score is stable and reliable. Runs evaluation multiple times and reports mean ± std.
- harness:deployRaphael ValdetaroUse when the user is done evolving and wants to finalize, clean up, tag the result, or push the optimized agent.
- harness:evolveRaphael ValdetaroUse when the user wants to run the optimization loop, improve agent performance, evolve the agent, or iterate on quality. Requires .evolver.json to exist (run harness:setup first).
- harness:healthRaphael ValdetaroUse when the user wants to check dataset quality, diagnose eval issues, or before running evolve. Checks size, difficulty distribution, dead examples, coverage, and splits. Auto-corrects issues found.
- harness:setupRaphael ValdetaroUse when the user wants to set up the evolver in their project, optimize an LLM agent, improve agent performance, or mentions evolver for the first time in a project without .evolver.json.
- harness:statusRaphael ValdetaroUse when the user asks about evolution progress, current scores, best version, how many iterations ran, or whether the loop is stagnating.