菜单
Run one improvement iteration. Chooses between PCO and IntelligentDesign based on recent effectiveness, runs the chosen approach, falls back to the other if it fails. Use when asked to "improve the agent", "run a coaching session", or "coach improve".
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Generate an HTML training dashboard from .coach/ state showing experiments, scores, learnings, TODOs. Pulls live tournament status. Opens in browser. Use for "dashboard", "training status", "show progress".
Save current coaching session state. Writes log, updates state.json and todos.md, commits and pushes if git is configured. Use at the end of each improvement cycle.
Run one full coaching step — start session, choose approach (PCO or IntelligentDesign), improve, save session. This is the skill that /loop calls every 30 minutes. Use when asked to "run a coaching step" or "coach step".
IntelligentDesign improvement iteration. Directly analyzes code and prompts, identifies a specific improvement, implements it, tests, and submits. Use when asked to "review the code", "improve the prompts", or "intelligent design".
PCO improvement iteration. Plays a game, collects experience, runs CvCLearner to propose program patches, tests locally, and submits if improved. Use when asked to "run PCO" or "evolve programs".
Start the coaching loop. Initializes state, then runs /coach.step every 30 minutes. Use when asked to "start coaching", "coach the agent", or "start the coaching loop".
| name | coach.improve |
| description | Run one improvement iteration. Chooses between PCO and IntelligentDesign based on recent effectiveness, runs the chosen approach, falls back to the other if it fails. Use when asked to "improve the agent", "run a coaching session", or "coach improve". |
Orchestrates improvement by choosing between /coach.improve.pco and /coach.improve.design based on effectiveness.
Assumes .coach/state.json and the current session directory already exist (via /coach.start-session). Reads IMPROVE.md for domain context.
Read .coach/state.json (check approach_stats) and .coach/todos.md. Pick the approach most likely to succeed this session.
{
"approach_stats": {
"pco": {"attempts": 5, "improvements": 2, "last_used": "20260330-..."},
"design": {"attempts": 8, "improvements": 5, "last_used": "20260331-..."}
}
}
Log the chosen approach in the session's plan.md.
Run /coach.improve.pco or /coach.improve.design.
If the chosen approach didn't produce an improvement (no accepted patch, or scores regressed):
After the session, update approach_stats in .coach/state.json:
stats = state["approach_stats"][approach]
stats["attempts"] += 1
if improved:
stats["improvements"] += 1
stats["last_used"] = session_timestamp
Also update .coach/todos.md with the approach tag:
- [x] (ID) Fixed chain-aware scoring in helpers/targeting.py — +41% avg
- [x] (PCO) Learner patched should_retreat with extra HP caution
- [x] (ID) Improved LLM prompt to return role suggestions