Agent Skills › PrimeIntellect-ai/prime-agent › refine

refine

GitHub

用于在Python REPL中触发持续工具集精炼。当观察到重复失败、可复用策略或行为规范时调用，以持久化更新提示词、记忆或技能配置。

packages/coding-agent/skills/refine/SKILL.md PrimeIntellect-ai/prime-agent

Trigger Scenarios

发现重复失败模式识别可复用战术确定需固化的行为策略

Install

npx skills add PrimeIntellect-ai/prime-agent --skill refine -g -y

More Options

Non-standard path

npx skills add https://github.com/PrimeIntellect-ai/prime-agent/tree/main/packages/coding-agent/skills/refine -g -y

Use without installing

npx skills use PrimeIntellect-ai/prime-agent@refine

指定 Agent (Claude Code)

npx skills add PrimeIntellect-ai/prime-agent --skill refine -a claude-code -g -y

安装 repo 全部 skill

npx skills add PrimeIntellect-ai/prime-agent --all -g -y

预览 repo 内 skill

npx skills add PrimeIntellect-ai/prime-agent --list

SKILL.md

Frontmatter

{
    "name": "refine",
    "description": "Trigger continual harness refinement from the Python REPL. Use when you notice a repeated failure, reusable tactic, delegation role, or behavior policy that should be persisted as a harness entry. Returns immediately; refinement runs when the current turn ends."
}

Refine

Refinement analyzes the conversation trajectory and applies small, evidence-backed updates to the continual harness (prompts, memories, skills, subagent specs). The implementation lives in the host (the same one behind the user's /refine command); this skill is the kernel-side interface to it. Call it directly from the Python REPL:

await refine.status()
await refine.run()
await refine.run("create a memory about always checking git status before committing")
await refine.run("promote the error-handling pattern to a global skill", global_=True)

API

await refine.status() — current refine state as a dict: pending (whether a requested refine is already queued for this turn) and in_flight (whether a refine is currently planning or applying).
await refine.run(instructions=None, global_=False) — schedule refinement. Returns {"scheduled": True} immediately, or {"scheduled": False, "reason": ...} when refinement cannot start. Optional instructions focus the refinement on a specific observation. Set global_=True to target the global harness store (cross-session); omit for local (session-scoped) refinement.

Rules

Refinement never runs mid-cell. A scheduled refinement runs when the current turn ends; the harness applies changes and rebuilds the system prompt, then resumes you automatically. Continue working normally after calling it.
One request per turn is enough; calling run again before the turn ends only updates the instructions.
Use refinement after observing a repeated failure, a reusable tactic, a repeated delegation role, or a behavior policy worth persisting. Do not rewrite the whole harness when a focused memory, skill, prompt note, or subagent spec is enough.