Token导航 LogoToken导航

Agent Skill · 研究检索

notion-brain(Notion brain 搜索)

用于处理 Notion 页面、数据库、工作区内容和结构化记录。它适合让 Agent 查询知识库、整理页面内容、创建记录或把外部信息同步到 Notion。使用时需要确认集成是否已被授权到目标页面或数据库,并区分读取、追加和覆盖更新;涉及批量写入或修改数据库属性时,应先核对字段名称、属性类型和目标页面。

notion-brain 是研究检索类 Agent Skill,可用宿主为 OpenClaw,通过ClawHub获取,权限为调用外部服务;许可证为 MIT-0,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 22,640。

总安装 4,499周安装 182GitHub Stars 公开资料未说明下载 1,412核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
MIT-0
版本
1.0.0

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:notion-brain(Notion brain 搜索)
来源仓库:https://github.com/zurbrick/notion-brain
安装命令:
openclaw skills install notion-brain
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 OpenClaw 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

ClawHubOpenClaw
openclaw skills install notion-brain

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

高质量内容路由到 Notion 工作区的智能网关。结合质量门控和目标映射精准投递信息。优化高价值内容的留存与使用效率。notion-brain 属于研究检索类 Skill,可作为该场景下的辅助能力补充。需定义质量评估标准和目标页面规则。写入模式可选择只读或运行时写入。

适用宿主包括 OpenClaw,接入前应确认版本、权限和运行环境要求。

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 OpenClaw;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 openclaw skills install notion-brain;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要调用外部服务,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

notion-brain 怎么安装?

在终端执行 openclaw skills install notion-brain。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

notion-brain 支持哪些 AI 工具?

公开资料标注支持 OpenClaw。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

notion-brain 需要什么权限?

权限级别是调用外部服务。该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

notion-brain 是免费的吗?

来源标注的许可证是 MIT-0。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

notion-brain 用的人多吗?

公开数据是累计安装 4,499 次,近一周安装 182 次,下载 1,412 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

notion-brain 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 专注 OpenClaw 一种宿主,触发方式和使用习惯不用来回切换。
  • 许可证明确为 MIT-0,商用和分发边界好判断。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 OpenClaw,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

不太适合

  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
notion-brain

Notion brain 搜索 · 本页

4,499—调用外部服务MIT-0通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

name
notion-brain
description
>

Notion Brain

Use this skill to make deliberate Notion saves, not reflexive ones.

The job is fourfold:

  1. decide whether the content deserves a durable home in Notion
  2. route it to the right page or database
  3. shape it into a clean structure before writing
  4. push it with the correct Notion MCP call pattern

Decision gate

Write to Notion only if at least one of these is true:

  • the content will be useful again later
  • it captures a decision, plan, status, insight, or reusable artifact
  • it would be annoying or costly to reconstruct
  • the owner would reasonably expect to find it in their second brain later

Do not write to Notion when the content is:

  • trivial, obvious, or disposable
  • already stored there in equivalent form
  • better suited only for chat, scratch work, or transient memory
  • part of a dedicated workflow this skill explicitly excludes

When unsure, prefer one of these outcomes:

  • Do not save
  • Save a short capture to Inbox DB
  • Save a fully structured page

Scope and non-scope

In scope:

  • research summaries
  • decision memos
  • project plans and project status
  • article drafts
  • security audit reports
  • financial snapshots
  • weekly rollups
  • contact notes
  • meeting prep
  • quick captures worth preserving

Out of scope:

  • health database management
  • property, vehicle, or equipment management
  • replacing workspace memory or daily logs
  • auto-pushing everything by default

Always write to workspace memory separately when the content also matters for agent continuity.

Default workflow

  1. Classify the content

Choose the closest content type from references/page-map.md.

  1. Choose save depth

- durable artifact or polished note -> structured page under the destination page - uncertain but worth keeping -> Inbox DB capture - existing page needs more detail -> append blocks instead of creating a duplicate

  1. Load the right template

Read references/templates.md and use the smallest template that preserves value.

  1. Check for an existing page first

Use Notion search before creating a new page when the topic may already exist.

  1. Write with MCP

Read references/mcp-commands.md and use the exact command pattern for: - search - create page in a parent page - create item in Inbox DB - update page metadata - append blocks to an existing page

  1. Keep titles specific

Prefer titles that are searchable and date-aware, for example: - Research Summary — Microsoft Copilot vs SecureAI — 2026-03-22 - Decision Memo — Notion Routing Skill - Weekly Rollup — 2026-W12

  1. Avoid junk writes

If the content is weak, incomplete, or duplicated, either tighten it first or do not save it.

Routing rules

Read references/page-map.md when deciding destination.

Practical defaults:

  • broad knowledge, synthesis, decisions, audits, rollups, and relationship notes -> Knowledge Hub
  • work execution and meeting support -> Work Hub
  • writing and publishing work -> Content Hub
  • money and finance artifacts -> Finance Hub
  • everything else worth saving but not yet sorted -> Inbox DB

Formatting rules

Read references/templates.md when preparing the payload.

Default formatting standards:

  • lead with a one-line summary
  • use short sections, not walls of text
  • preserve source links, dates, and decisions
  • include Next steps only when action is implied
  • avoid raw dump formatting unless using Inbox DB for quick capture

MCP usage

Read references/mcp-commands.md before writing. Use native Notion MCP tools (notion-search, notion-create-pages, notion-update-page, notion-fetch) — not mcporter.

Command order:

  1. search if collision is possible
  2. create or identify target page/database item
  3. add content (inline for new pages, append for existing)
  4. update page properties only when needed

Duplicate handling

Always search before creating when:

  • the title contains a person name, project name, or recurring topic
  • the content type recurs (weekly rollup, project status, financial snapshot)
  • you are unsure whether the page already exists

When a duplicate is found:

  • rollups and status pages → use replace_content to supersede the old version
  • knowledge pages → use update_content to append new findings
  • captures → skip the write and note the existing page

Content size limits

Notion API constrains each rich text element to ~2000 characters. For content longer than ~1500 words, split into multiple calls. Keep each payload focused; add long sections via follow-up update_content calls.

Review markers

After pushing content to Notion, consider adding a comment instead of editing the page body when:

  • marking that the content was reviewed or verified
  • flagging something as stale
  • adding metadata about when/why the page was created

Use the Comments API pattern from references/mcp-commands.md.

Setup

Before using this skill, replace the placeholder page and database IDs in the reference files with your own Notion IDs.

Minimum setup:

  • references/page-map.md — set the destinations you want to use for knowledge, work, content, finance, and inbox capture
  • references/mcp-commands.md — replace every YOUR_*_ID placeholder with the matching Notion page or data source ID
  • optional: rename destination labels to match your workspace, while preserving the routing logic

Recommended destination mapping:

  • Knowledge Hub — durable notes, research, decisions, audits, rollups, relationship notes
  • Work Hub — active execution artifacts, meeting prep, project plans, project status
  • Content Hub — drafts, outlines, posts, newsletters, content experiments
  • Finance Hub — durable financial summaries and snapshots
  • Inbox DB — quick captures worth saving before full structuring

How to get your IDs:

  • open the target page or database in Notion
  • copy the page URL
  • extract the 32-character page/database identifier from the URL
  • use the page ID for top-level pages and the data source ID for databases such as Inbox DB

Do not ship or publish your private Notion IDs in shared repositories.

Output expectations

When using this skill, return a compact operator-style summary:

  • Decision: save / do not save / save to inbox
  • Destination: exact Notion page or database
  • Format: template used
  • Action: command pattern chosen or write completed
  • Memory note: whether the same content should also be logged to workspace memory

References

  • references/page-map.md — content type to destination routing
  • references/templates.md — section templates by content type
  • references/mcp-commands.md — exact Notion MCP command patterns with placeholder IDs

平台分布

OpenClaw

73.56%
按下载量换算1,039

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

通过

权限和风险

调用外部服务

该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills