菜单
Run an evolution checkpoint — recall relevant past outcomes, reflect on the current task, and record what was learned. Use when the user requests an evolution checkpoint or asks to "evolve" or "checkpoint".
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。
Show Evolver health — evolution memory location & size, recent log, and whether the full engine is installed. Use when the user asks for "evolver status", "evolver health", or "check evolver".
Self-evolution workflow for the agent. Before a substantive task, recall what worked on similar past tasks from evolution memory; after it, record the outcome so future sessions learn from it. Use when the user starts non-trivial work (a feature, a fix, a refactor) or asks the agent to "evolve", "learn from this", or "remember how this went".
Distill a reusable skill/gene from recent run history (optionally from an LLM response file). Use when the user asks to "distill skill", "extract gene", or "generalize lessons".
Review Evolver's pending evolved changes, then approve (solidify) or reject (roll back). Use when the user asks to "review changes", "accept evolver changes", or "discard/rollback evolver changes".
Run one Evolver self-evolution cycle on the current repo (collect signals → select/mutate genes → propose changes). Use when the user asks to "run evolution cycle", "start evolver loop", or "run evolver strategy".
Solidify the current working changes into a durable Evolver gene/capsule (with rollback safety). Use when the user requests to "solidify changes", "save gene", or "capture this capsule".
| name | evolver-evolve |
| description | Run an evolution checkpoint — recall relevant past outcomes, reflect on the current task, and record what was learned. Use when the user requests an evolution checkpoint or asks to "evolve" or "checkpoint". |
Execute a manual checkpoint in the evolution lifecycle of the current task.
When triggered, guide yourself through these three steps:
Recall. Look at the evolution memory the SessionStart hook injected (or read the tail of the memory graph at ~/.evolver/memory/evolution/memory_graph.jsonl, or the project's local memory/evolution/memory_graph.jsonl if present). Summarize any recent outcome — success or failure — that is relevant to what is currently being worked on.
Reflect. Given the current diff or task state, formulate a concise reflection:
Record. Remind the user that the Stop hook records outcomes automatically at task/session end. If they want to run the full engine now, and @evomap/evolver is on PATH, run:
evolver run
If it is not installed, inform them that the outcome will still be captured automatically by the Stop hook, and that running npm install -g @evomap/evolver unlocks the full review-and-solidify cycle.
Keep the process lightweight and focused on generating a durable lesson.