ratchet-review

仓库创建 2026年6月30日最近提交 21 天前SkillHot 收录 21 天前

▸ 精选理由

保证知识库结论经人审查，适用于治理与质量控制流程。

这个 Skill 做什么

为 ratchet 提供人工复核门，将待定结论接受为概念或拒绝并调整结构。

把 ratchet 抓取出的“claims”（待定结论）做人工复核，把成熟的接受为概念（concepts），或拒绝并调整其提议的结构。通常在需要把自动挖掘的学习成果人工把关、整理进知识库时用。它是双层的人类门控，重点做可信度和结构化的核查，最终由 sulin 做决策。

▸ 展开 SKILL.md 英文原文

Review ratchet's pending claims (né takeaways) and queued structural-op proposals — the human review gate (two tiers). Invoke when sulin wants to triage what ratchet has learned from his sessions, or accept/reject the gardener's proposed concept restructurings. You are an active faithfulness-checker; sulin makes every call.

垂直行业审查流程知识治理ratchetClaude Code

Stars

Forks

仓库内 Skill

7 日增星

安装 / 使用

给你的 Agent 一句话（通用）

帮我安装这个 skill：https://raw.githubusercontent.com/SteveUlin/ratchet/main/.claude/skills/ratchet-review/SKILL.md

或 curl 直取 SKILL.md

curl -fsSL "https://raw.githubusercontent.com/SteveUlin/ratchet/main/.claude/skills/ratchet-review/SKILL.md"

SKILL.MD 节选查看完整文件 ↗

# ratchet-review — the human gate

ratchet mines sulin's Claude Code sessions into **claims** (a lesson that recurred, with verified evidence and — once matured — a synthesized "why"). This skill is the one hard gate, and it has **two tiers**:

- **Tier 1 — claims → concepts** (most of this skill): a claim sulin **accepts** becomes a **concept** (ratchet's curated knowledge), which feeds back into the system and is later projected into CLAUDE.md/skills.
- **Tier 2 — structural-op proposals** (see the section at the end): the *gardener* proposes restructurings of the concept layer itself — merge two concepts that say the same thing, split one that conflates two lessons, retire a stale one. Th

via SKILL·HOT · 数据来自 GitHub 公开信息 · 原文版权归作者所有