Claude Skill

blader/theorist

Theorist 是一个 Codex/Claude Skill,自动维护每个仓库的操作理论文档,确保 AI 助手与项目约定保持一致。

概览

Stars177
Forks7
语言未知
最后更新2026-02-28
最近同步2026-06-23
前往 GitHub

仓库信息

拥有者blader
仓库theorist
完整名称blader/theorist
Repo ID1,169,534,442

安装这个 Skill

git clone https://github.com/blader/theorist.git "$SKILLS_DIR/theorist"

Registry 信息

类型openclaw_skill
质量分70/100
验证状态readme_parsed
最近验证2026-06-23
平台
ClaudeCodex
能力
pdfmemoryterminalworkflow
识别文件
README.mdSKILL.md

项目简介

Theorist 是一个 Codex/Claude Skill,自动维护每个仓库的操作理论文档,帮助 AI 助手理解项目上下文和约定。

英文描述

A Codex/Claude skill that maintains per-repo operating theory documents.

要点

  • 自动生成并更新每个仓库的理论文档
  • 与 Codex 和 Claude 工作流集成
  • 保持 AI 上下文与项目约定一致
  • 轻量且无侵入的配置方式

使用场景

  • 让 AI 快速了解新仓库
  • 跨项目保持 AI 行为一致性
  • 记录项目特定的编码约定
  • 减少 AI 助手的上下文切换开销

README 摘要

# Theorist Maintain a continuously updated operating theory for ongoing engineering work. `theorist` is an agent skill (Codex and Claude compatible) that keeps a per-repo narrative document at `THEORY.MD` (repo root). The document explains the problem thesis, the current system model, why the strategy is shaped the way it is, and where uncertainty still exists. It is intentionally not a task plan, changelog, or status report. ## Install Use a skills directory for your client and clone into it: ```bash SKILLS_DIR="${CODEX_SKILLS_DIR:-${CLAUDE_SKILLS_DIR:-$HOME/.codex/skills}}" mkdir -p "$SKILLS_DIR" git clone https://github.com/blader/theorist.git "$SKILLS_DIR/theorist" ``` If you already have the repo checked out, copy just the skill file: ```bash mkdir -p "$SKILLS_DIR/theorist" cp SKILL.md "$SKILLS_DIR/theorist/" ``` ## Usage Invoke it using your client's skill command. Or ask naturally: - "Keep the operating theory updated while you work." - "Rewrite the theory doc based on what we just learned." - "What changed in our theory after this debugging pivot?" ## What it enforces - One theory doc per repo at `THEORY.MD` (repo root) - Always-on activation in every session - Sticky activation: once active, it stays active for the full session - Holistic rewrites (not appended logs) - Clear separation between: - problem thesis, - operating theory, - strategy, - key discoveries/pivots, - open questions. - Frequent in-session updates as learnings arrive (especially after investigate/implement/verify loops and verification outcomes). - Updates when understanding changes, not when code churn happens. ## Why this helps Most agent workflows optimize for local task completion. `theorist` adds a lightweight strategic memory layer so future decisions remain

话题

暂无话题

探索更多

相关技能

你可能也会感兴趣的 Claude Skill 项目。

数据来自 GitHub,同步时间:2026-06-23