一键导入
wiki
Maintain and query a personal LLM Wiki — an incremental, compiled knowledge base of structured Markdown files.
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。
菜单
Maintain and query a personal LLM Wiki — an incremental, compiled knowledge base of structured Markdown files.
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
Planner-Generator-Evaluator 编排模式。GAN 启发的多 agent 动态编排,通过 ai CLI 控制子 agent 完成复杂任务的拆解-执行-验证闭环。
Code review skill using codex-rs methodology with ai CLI
为任意代码仓库构建结构化知识库(code wiki),并通过 frozen snapshot 机制启动即问即答的 expert agent。 当你需要深入理解一个大型代码仓库、构建可复用的代码知识库、或为其他 agent 提供代码专家服务时使用此技能。
Use the mcporter CLI to list, configure, auth, and call MCP servers/tools directly (HTTP or stdio), including ad-hoc servers, config edits, and CLI/type generation.
Orchestrate a real-time alternating debate between two subagents using ai serve/send. Judge controls rounds by prompting each agent in turn.
任务规划与进度追踪。当任务有 5+ 步骤、多文件改动、或长执行链时使用。 核心机制:用 checkbox 列表追踪进度,每完成一步立即更新,利用"重写列表"将完整任务清单重新注入 context。
| name | wiki |
| description | Maintain and query a personal LLM Wiki — an incremental, compiled knowledge base of structured Markdown files. |
| metadata | {"goclaw":{"emoji":"📚","requires":{}}} |
An LLM-maintained wiki that compiles knowledge once and keeps it current, instead of re-deriving from raw sources on every query.
Default: ~/project/zen/
This is an Obsidian vault. The wiki lives at:
~/project/zen/
├── AGENTS.md # Wiki schema & conventions
├── raw/ # Immutable source documents
│ ├── webclips/
│ ├── papers/
│ ├── repos/
│ ├── chat/ # Curated AI chat transcripts with learning value
│ └── assets/
├── wiki/
│ ├── index.md # Hub — lists all pages, updated on every ingest
│ ├── log.md # Append-only activity log
│ ├── summaries/ # One summary page per source
│ ├── concepts/ # Entity / topic pages (evolving synthesis)
│ └── entities/ # Entity pages (reserved)
If the user's project has a different vault path, adjust accordingly.
When the user provides a new source (file, URL, pasted text):
raw/webclips/ (or raw/papers/, raw/repos/ as appropriate). Filename: slug-name.md.wiki/summaries/slug-name.md:
tags, date, sources## Related Notes with [[wikilinks]]wiki/concepts/:
wiki/index.md — add new entries under the correct categorywiki/log.md — format:
## [YYYY-MM-DD] ingest | Source Title
- **Sources**: path to raw file(s)
- **Created**: list of new wiki pages
- **Updated**: list of updated wiki pages
- **Notes**: brief description
When the user asks a question about knowledge in the wiki:
wiki/index.md to locate relevant pagesPeriodically or on request:
[[wikilinks]]index.md still existEvery wiki page must have:
---
tags: [relevant, tags]
date: YYYY-MM-DD
sources: [source-slug]
---
Every wiki page should have:
## Related Notes section with [[wikilinks]]lowercase-with-hyphens.md[[filename-without-ext]] (Obsidian convention)At current scale (~dozens of pages):
wiki/index.md is sufficient for locating pageswiki/scripts/wiki-search.sh <query> for content search (uses rg if available)wiki/scripts/wiki-search.sh <query> --raw to also search raw/ sourcesAs the wiki grows (>100 pages):
npm install -g @tobilu/qmd (requires Node.js)Run wiki/scripts/wiki-lint.sh to verify:
Exit code = error count (0 = healthy).
| Action | How |
|---|---|
| Add a source | Ingest workflow (5 steps above) |
| Ask a question | Read index → read pages → synthesize |
| Find a page | Check wiki/index.md |
| Search content | wiki/scripts/wiki-search.sh "query" |
| Search including raw | wiki/scripts/wiki-search.sh "query" --raw |
| Health check | wiki/scripts/wiki-lint.sh |
| Browse visually | Open Obsidian → Graph View |
| Check recent activity | grep "^## \[" wiki/log.md | tail -5 |
| Check wiki health | Verify index ↔ actual files match |