Synapse Wiki

v2.0.1

Synapse Wiki — 智能知识库管理系统。 自动摄取原始资料,增量构建持久化知识网络,支持智能查询和健康检查。 知识随时间复利积累,越用越聪明。 当用户提到 wiki、知识库、摄取资料、查询知识、整理文档时使用此技能。

0· 225·1 current·1 all-time
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
Purpose & Capability
Name/description (Synapse Wiki) align with the included files and behavior: scaffold/ingest/query/lint scripts, shell wrappers, tests, and documentation. Required binaries are limited to python3, which is appropriate.
Instruction Scope
SKILL.md and the command scripts only read and write files inside the user-specified wiki root and the skill install location. Ingest enforces a path check so source files must be inside the wiki root. There are no instructions to read unrelated system files, environment secrets, or to send data to external endpoints.
Install Mechanism
There is no remote download/install of third-party code; install.sh copies the skill into $HOME/.claude/skills and creates a config.json from a template. This is a local file operation (expected) but it will overwrite an existing installation when run without --dry-run; review the script before running and consider using --dry-run first.
Credentials
The skill requests no environment variables or credentials. Scripts operate on filesystem paths provided by the user and do not access network credentials, cloud keys, or unrelated environment data.
Persistence & Privilege
always:false and user-invocable:true (normal). The skill can be invoked autonomously by the model (disable-model-invocation:false), which is the platform default. The SKILL.md also documents optional integration with a 'synapse-brain' orchestrator — this is an architectural integration, not a credentials/privilege request. Consider that autonomous invocation combined with local write access means the agent could run ingest/lint/query actions on user-provided wiki roots if triggered.
Assessment
This skill appears to do what it says: local file-based wiki scaffolding, ingestion, query synthesis, and linting using Python scripts. Before installing: 1) Inspect install.sh and run it with --dry-run to avoid accidental overwrites; back up ~/.claude/skills if present. 2) Only point the tool at directories you intend to manage (don't set wiki_root to system or sensitive directories). Ingested files and generated outputs live under the wiki root and will be read by the LLM during queries — avoid ingesting secrets or confidential files. 3) If you don't want the agent to autonomously run ingest/lint on session start, keep model-driven invocation disabled in agent settings or avoid enabling Brain scheduling. 4) If you expect remote integrations, review or add explicit network/credential handling (none is present here). Overall the package is internally consistent and local-only; treat it like any third-party script and run it in a controlled environment first.

Like a lobster shell, security has layers — review code before you run it.

Runtime requirements

🧠 Clawdis
Binspython3
latestvk975a40rjfs4dqtvd0dv2e6bsh84w2r5
225downloads
0stars
10versions
Updated 1w ago
v2.0.1
MIT-0

Synapse Wiki Skill

Synapse Wiki = 持久化知识网络 + 增量积累 + 自动健康维护

核心理念:知识应该随时间复利增长,而非每次从零开始。

传统笔记Synapse Wiki
机制手动整理 + 检索自动编译原始资料成知识网络
知识形式孤立文档结构化、交叉链接的知识页面
随时间变化越积越多,难以查找越用越聪明,自动关联
维护者人类AI(编译、交叉引用、归档)

🚦 快速决策:我该用什么命令?

你想做什么?
│
├─ 开始一个新知识库         → ingest(初始化)
│   └─ 例:"我想建一个 AI 学习知识库"
│
├─ 保存新资料/文章          → ingest(摄取)
│   └─ 例:"这篇好文要存起来"、"剪辑网页保存"
│
├─ 查询已有知识             → query(查询)
│   └─ 例:"RAG 是什么?"、"上次那个概念怎么解释"
│
└─ 检查知识库健康度         → lint(检查)
    └─ 例:"有没有死链接?"、"检查孤立页面"

常用场景:

  • 第一次用 → /synapse-wiki init ~/my-wiki "AI 知识库"
  • 看到好文章 → /synapse-wiki ingest ~/my-wiki raw/articles/xxx.md
  • 有疑问 → /synapse-wiki query ~/my-wiki "你的问题"

📋 命令速查卡片

命令用途示例
/synapse-wiki init初始化知识库/synapse-wiki init ~/my-wiki "AI 知识库"
/synapse-wiki ingest摄取新资料/synapse-wiki ingest ~/my-wiki raw/articles/article.md
/synapse-wiki query查询知识/synapse-wiki query ~/my-wiki "RAG 是什么"
/synapse-wiki lint健康检查/synapse-wiki lint ~/my-wiki

使用提示:

  • 首次使用先用 init 创建目录结构
  • 每次保存资料后用 ingest 编译为知识
  • 定期运行 lint 检查健康度

三层架构

<wiki-root>/
├── CLAUDE.md              ← Schema 定义(范围/规范/工作流)
├── log.md                 ← 只增不减的时间线日志
│
├── raw/                   ← 原始资料层(LLM 只读,永不修改)
│   ├── articles/          ← 网页文章(Obsidian Clipper 保存)
│   ├── papers/            ← 学术论文
│   └── notes/             ← 个人笔记
│
└── wiki/                  ← Wiki 知识层(LLM 编写,用户阅读)
    ├── index.md           ← 主目录:所有页面 + 一句话摘要
    ├── concepts/          ← 概念/主题页面
    ├── entities/          ← 人物、工具、论文、组织
    └── summaries/         ← 每个来源的摘要页面

命令

摄取命令

# 摄取新资料
/synapse-wiki ingest /path/to/wiki "raw/articles/article.md"

查询命令

# 查询 Wiki 知识
/synapse-wiki query /path/to/wiki "LLM Wiki 的核心思想"

健康检查命令

# Wiki 健康检查
/synapse-wiki lint /path/to/wiki

初始化命令

# 初始化新的 Wiki 知识库
/synapse-wiki init /path/to/wiki "AI 学习知识库"

Scripts

脚本用途
scripts/scaffold.py引导新的 Wiki 目录树
scripts/ingest.py摄取新资料,编译为 Wiki 页面
scripts/query.py查询 Wiki,综合答案
scripts/lint_wiki.py健康检查(死链接/孤立页/矛盾)

页面类型模板

概念页面 (400-1200 词)

---
title: <Title>
type: concept
created: YYYY-MM-DD
updated: YYYY-MM-DD
sources: [slug1, slug2]
tags: [tag1, tag2]
---

# <Title>

<一句话定义或核心思想。>

## What it is
<清晰解释。>

## How it works
<机制、过程或结构。>

## Key properties / tradeoffs
<重要特征。>

## Relationship to other concepts
<相关概念的 wikilinks。>

## Open questions
<Wiki 尚未解决的问题。>

实体页面 (200-500 词)

---
title: <Name>
type: entity
entity_type: person | tool | paper | organization | project
created: YYYY-MM-DD
updated: YYYY-MM-DD
sources: [slug1]
tags: [tag1]
---

# <Name>

<一句话描述。>

## Key contributions / features
<主要贡献或特性。>

## Related concepts
<相关概念。>

Summary 页面 (150-400 词)

---
title: summaries/<slug>
type: summary
source_type: article
date: YYYY-MM-DD
ingested: YYYY-MM-DD
tags: []
---

# <Source Title>

## Key takeaways
- <最重要洞察 1>
- <最重要洞察 2>
- <最重要洞察 3>

## Core claims
<主要论点的 2-4 句话摘要。>

## Concepts introduced / referenced
<概念和实体。>

索引和日志

index.md 格式

# Index — <Wiki Name>

## Concepts(概念)
- [[Page Name]] — One-line summary

## Entities(实体)
- [[Page Name]] — One-line summary

## Summaries(资料摘要)
- [[summaries/slug]] — One-line summary

log.md 格式

## [YYYY-MM-DD] ingest | <slug> — <description>
## [YYYY-MM-DD] query | <question-slug>
## [YYYY-MM-DD] lint | <N> issues found
## [YYYY-MM-DD] promote | <page-name> (from query)

Session 启动检查清单

每个新 Session:

  1. 读取 CLAUDE.md,确认范围和规范
  2. 读取 log.md 最近 5 条:grep "^## \[" log.md | tail -5
  3. 如有新 raw/ 资料,执行 Ingest
  4. 如用户提问,执行 Query(先查 index.md)
  5. 如 ingest 超过 10 次未 lint,执行 Lint

Brain 调度集成 (v2.0 新增)

Synapse Wiki v2.0 可作为 synapse-brain 的被调度 Hand Agent 运行:

用户 → synapse-brain(意图识别:知识查询/保存)
         ↓ 路由到 wiki
       synapse-wiki(执行 ingest/query/lint)
         ↓ 完成后
       synapse-brain(汇总 + 状态保存)

通过 Brain 调度使用:

/synapse-brain dispatch "保存这篇 RAG 文章" --skill synapse-wiki
/synapse-brain dispatch "RAG 是什么?" --skill synapse-wiki

与 synapse-code 互操作

synapse-code Pipeline 完成后可自动触发 wiki 知识沉淀:

synapse-code Pipeline 完成
         ↓ (auto_log enabled)
    synapse-wiki ingest
         ↓
    知识写入 wiki/summaries/
         ↓
    包括:技术决策、代码模式、Bug 修复记录

配置方式:

{
  "interop": {
    "wiki_enabled": true,
    "wiki_root": "~/my-project/wiki"
  }
}

知识沉淀内容:

  • 技术决策 — 为什么选择方案 A 而非 B
  • 代码模式 — 项目中常用的架构模式
  • Bug 修复 — 遇到的坑和解决方案
  • API 变更 — 接口设计决策和版本记录

安装

# 方式 1: 使用安装脚本(推荐)
cd ~/.claude/skills/synapse-wiki
./install.sh

# 方式 2: 手动复制
cp -r synapse-wiki ~/.claude/skills/

# 方式 3: OpenClaw (如有 .skill 文件)
claude skill install synapse-wiki.skill

使用场景

  • 📚 个人知识库建设 — 积累 AI/技术知识,构建第二大脑
  • 📝 项目文档管理 — 维护项目 Wiki,团队成员快速上手
  • 🔍 知识检索 — 快速查找已学概念,不再大海捞针
  • 🧹 知识整理 — 定期健康检查,保持知识网络整洁

Comments

Loading comments...