← All skills
harness:evolve
80 / 100 · Low risk
Fine to try. Some of what it can do is expected: a coding skill usually runs build and test commands.
By raphaelchristi/harness-evolver (GitHub) · coding · 57 stars · scanned at commit 87fa761
Use 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).
Works with any agent that supports SKILL.md skills.
What this skill can do
Runs commands on your computer
The author does not mention this. It runs automatically when the plugin is installed. Expected for a coding skill.
Found in 20 places.
Show 17 more
- skills/evolve/SKILL.md, line 85
BEST=$(python3 -c "import json; b=json.load(open('.evolver.json')).get('best_experiment'); print(b if b else '')") - skills/evolve/SKILL.md, line 92
ITER_TRACE=$($EVOLVER_PY $TOOLS/log_iteration.py --config .evolver.json --action start --version v{NNN} 2>/dev/null) - skills/evolve/SKILL.md, line 106
if [ -n "$BEST" ]; then - skills/evolve/SKILL.md, line 134
SRC="$(dirname "$(git rev-parse --git-common-dir)")" - skills/evolve/SKILL.md, line 178
CONCURRENCY=$(python3 -c "m={'light':5,'balanced':3,'heavy':3}; print(m.get('$MODE',3))") - skills/evolve/SKILL.md, line 195
LLM_EVALS=$(python3 -c "import json; c=json.load(open('.evolver.json')); llm=[k for k in c['evaluators'] if k in ('correctness','conciseness')]; print(','.join( - skills/evolve/SKILL.md, line 212
$EVOLVER_PY $TOOLS/read_results.py --experiments "{names}" --config .evolver.json --split held_out --output comparison.json - skills/evolve/SKILL.md, line 219
$EVOLVER_PY $TOOLS/read_results.py --pairwise "{winner},{runner_up}" --config .evolver.json --split held_out - skills/evolve/SKILL.md, line 225
WINNER_PROJECT="{winner_wt}" - skills/evolve/SKILL.md, line 241
$EVOLVER_PY $TOOLS/update_config.py --config .evolver.json --action backup - skills/evolve/SKILL.md, line 261
git tag "evo-iter-v{NNN}" -m "harness: v{NNN} score={score}" - skills/evolve/SKILL.md, line 270
for CANDIDATE in {all_worktree_paths}; do - skills/evolve/SKILL.md, line 277
$EVOLVER_PY $TOOLS/regression_tracker.py --config .evolver.json --previous-experiment "$PREV" --current-experiment "$WINNER" --add-guards --auto-guard-failures - skills/evolve/SKILL.md, line 284
ITER_DURATION=$(( $(date +%s) - ITER_START )) - skills/evolve/SKILL.md, line 308
$EVOLVER_PY $TOOLS/cleanup_worktrees.py --dir "$SRC" - skills/evolve/SKILL.md, line 322
$EVOLVER_PY $TOOLS/evolution_chart.py --config .evolver.json - hooks/hooks.json, line 9
"command": "bash \"${CLAUDE_PLUGIN_ROOT}/hooks/session-start.sh\""
Downloads and installs other software
The author does not mention this. It runs automatically when the plugin is installed.
Found in 3 places.
Install
Paste this into your terminal. It downloads exactly the version we scanned, not whatever is newest.
if [ -e ~/.agents/skills/'harness-evolve' ] || [ -L ~/.agents/skills/'harness-evolve' ]; then echo 'A skill folder named harness-evolve already exists — not replacing it' >&2; false; else d=$(mktemp -d) && git clone --quiet --filter=blob:none --no-checkout 'https://github.com/raphaelchristi/harness-evolver' "$d" && git -C "$d" checkout --quiet '87fa7612358acccb01d34abf72426a7e47329642' -- 'skills/evolve' && mkdir -p ~/.agents/skills && cp -R "$d"/'skills/evolve' ~/.agents/skills/'harness-evolve' && rm -rf "$d"; fi
Codex reads ~/.agents/skills directly. For Claude Code, Cursor, Copilot, Gemini CLI, Antigravity or OpenCode, also link it. For Claude Code: ln -s ~/.agents/skills/'harness-evolve' ~/.claude/skills/'harness-evolve' (other agents: ~/.cursor/skills, ~/.copilot/skills, ~/.gemini/skills, ~/.gemini/config/skills, ~/.config/opencode/skills). Or install through the finder skill, which does both.
Badge for your README
The badge for raphaelchristi/harness-evolver shows the score of its lowest-scoring skill.
[](https://zalavijaysinh.github.io/skilltrust/repos/raphaelchristi/harness-evolver/)
Score history
| Date | Version | Score |
|---|
| 2026-11-07 | 87fa761 | 80 |