Token导航 LogoToken导航

Agent Skill · 研究检索

second-brain(第二大脑)

second-brain 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

second-brain 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为接触敏感数据;许可证为 MIT,第三方扫描有未通过项,按安装量在研究检索类的 73,675 个 Skill 里排第 25,008。

总安装 3,868周安装 158GitHub Stars 27下载 1,251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

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:second-brain(第二大脑)
来源仓库:https://github.com/sean-esk/second-brain-gtd
仓库路径:skills/second-brain
安装命令:
npx skills add https://github.com/sean-esk/second-brain-gtd --skill second-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。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/sean-esk/second-brain-gtd --skill second-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。

second-brain 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词快速定位候选结果时使用。它可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网或文件读写。

适用于研究检索类任务,如知识管理与信息搜集。通过 npx skills add 命令从指定 GitHub 仓库安装。

来源仓库内路径:skills/second-brain

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 Codex、Claude、Cursor、Gemini CLI;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 npx skills add https://github.com/sean-esk/second-brain-gtd --skill second-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

    它需要接触敏感数据,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

second-brain 怎么安装?

在终端执行 npx skills add https://github.com/sean-esk/second-brain-gtd --skill second-brain。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

second-brain 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。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。

second-brain 需要什么权限?

权限级别是接触敏感数据。该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

second-brain 是免费的吗?

来源标注的许可证是 MIT。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。

second-brain 用的人多吗?

公开数据是累计安装 3,868 次,近一周安装 158 次,下载 1,251 次,GitHub 27 星。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。

second-brain 安全吗?

第三方扫描有未通过项,风险较高。除非你确认过源码,否则不建议直接安装到常用环境。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 3 个目录同时收录,说明这个 Skill 流传比较广。
  • 许可证明确为 MIT,商用和分发边界好判断。

先想清楚的地方

  • 需要接触敏感数据,装到常用环境前要先评估影响面。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 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 Codex,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。

不太适合

  • 环境里有大量密钥、凭证,不方便给第三方 Skill 读取的人。
  • 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
second-brain

第二大脑 · 本页

3,86827接触敏感数据MIT未通过

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

Second Brain

A personal knowledge management and productivity system for Obsidian, combining:

  • GTD (Getting Things Done) - Task and project management
  • Zettelkasten - Atomic note-taking for knowledge building
  • PARA Method - Folder organization (Projects, Areas, Resources, Archives)

Configuration (MEMORY-BASED)

CRITICAL FIRST STEP: Before any operation, check for configuration in Claude's Memory.

How Configuration Works

Configuration is stored in Claude's Memory feature and persists across all sessions automatically.

Why Memory? Skills in Claude Desktop run in a sandboxed environment with no file system access. Memory is the only way to persist configuration across sessions.

Once set up, Claude remembers:

  • Your vault path
  • Your name
  • Whether setup is complete
  • Your preferences

Configuration Check Flow

  1. Check Claude Memory for "Second Brain vault path"
  2. If found in Memory: Use the remembered vault path for all operations
  3. If NOT found in Memory: This is first-time setup - ask user for vault path

Checking Memory

At the start of ANY Second Brain operation, check if you remember the vault path:

Do I have the user's Second Brain vault path in my memory?
- If YES: Use that path for all operations
- If NO: Ask the user and save to memory

If No Configuration Found (First Time)

Ask the user for their vault path:

Welcome to Second Brain!

I need to know where your Obsidian vault is located so I can save and organize your notes.

Please provide the full path to your Obsidian vault folder.

Examples:
- macOS: /Users/yourname/Documents/MyVault
- Windows: C:\Users\yourname\Documents\MyVault
- Linux: /home/yourname/Documents/MyVault

What's the path to your vault?

Once you have the path:

  1. Save to Claude Memory: Remember the vault path for future sessions
  2. Run the full setup workflow to complete onboarding
  3. Save setup completion status to Memory

Claude Memory persists across ALL sessions automatically.

What Gets Saved to Memory

After setup, Claude should remember:

  • Second Brain vault path: The absolute path to the user's Obsidian vault
  • Second Brain user name: The user's name
  • Second Brain setup complete: Whether full setup has been done
  • Second Brain preferences: User preferences (proactive capture, inbox threshold)

If Configuration Exists in Memory

Use the remembered vault path for ALL file operations.

Claude Code Fallback

For Claude Code users only: If Memory is empty, you may also check for a legacy config file at ~/.second-brain/config.json. If found, migrate that configuration to Memory for future use.

Environment Variable Override

Environment variable SECOND_BRAIN_VAULT_PATH can override Memory (useful for testing).


Core Capabilities

1. Quick Capture

Trigger phrases: "capture this", "save this thought", "remember this", "note this down", "add to my inbox"

What it does: Instantly captures thoughts, tasks, or ideas to today's inbox without categorization.

GTD Principle: "Capture first, clarify later."

See: Capture Workflow


2. Daily Planning

Trigger phrases: "plan my day", "what should I work on", "daily planning", "morning planning"

What it does:

  • Checks inbox (prompts to process if 5+ items)
  • Scans all projects for next actions
  • Asks about energy, context, time available
  • Generates prioritized task list (must-do, should-do, quick-wins)

See: Daily Plan Workflow


3. Inbox Processing

Trigger phrases: "process my inbox", "organize my captures", "clarify my tasks", "GTD processing"

What it does:

  • Asks clarifying questions for vague items (batched, not one-by-one)
  • Routes actionable items to Projects/Areas
  • Routes knowledge items to Permanent Notes
  • Reviews all active projects
  • Archives completed projects

See: Process Inbox Workflow


4. Daily Closeout

Trigger phrases: "daily closeout", "review my day", "end of day review", "evening reflection"

What it does:

  • Reads today's plan
  • Asks what was accomplished
  • Marks completed/partial/deferred items
  • Asks about tomorrow's priorities
  • Creates tomorrow's DRAFT plan

See: Daily Closeout Workflow


5. Setup & Configuration

Trigger phrases: "set up my second brain", "configure my vault", "second brain setup", "reconfigure"

What it does:

  • First-time: Full interactive setup (goals, relationships, first project)
  • Re-run: Update existing configuration

See: Setup Workflow


6. Excalidraw Diagrams

Trigger phrases: "create a diagram", "draw a flowchart", "make an excalidraw", "visualize this", "sketch this out", "diagram showing"

What it does:

  • Creates .excalidraw.md files compatible with Obsidian Excalidraw plugin
  • Generates flowcharts, concept maps, system diagrams, mind maps
  • Supports shapes (rectangles, ellipses, diamonds), arrows, lines, and text
  • Full JSON structure with proper element properties

Output location: {{vaultPath}}/ (user specifies location or defaults to appropriate folder)

See: Excalidraw Reference


Proactive Capture

IMPORTANT: When you notice the user:

  • Discovering valuable information during research
  • Having insights or realizations
  • Discussing ideas worth preserving
  • Solving problems in interesting ways

Offer to capture:

That's an interesting insight about [topic]. Would you like me to capture this to your Second Brain?

If they agree:

  • Use the capture workflow
  • Add context about where the insight came from
  • Suggest relevant tags or connections

Vault Structure

The system uses PARA + Zettelkasten organization:

{{vaultPath}}/
├── 00-Inbox/
│   ├── Daily/              # Captures go here (YYYY-MM-DD.md)
│   └── Fleeting-Notes/     # Knowledge items during processing
├── 01-Projects/            # Multi-step outcomes with deadlines
├── 02-Areas/               # Ongoing responsibilities
│   ├── Career-Development.md
│   ├── Health-Fitness.md
│   ├── Personal-Development.md
│   ├── Errands.md
│   ├── Personal-Todos.md
│   └── Relationships/      # Individual notes per person
├── 03-Resources/
│   └── Reference-Notes/    # Summaries of external sources
├── 04-Archives/            # Completed/inactive projects
├── Daily Plans/            # Generated daily plans
├── Meeting Notes/          # Meeting documentation
├── Permanent Notes/        # Zettelkasten - synthesized insights
│   └── Assisting-User-Context.md  # User's goals & context
└── Templates/              # Reusable templates

See: PARA + Zettelkasten Guide


Unified Task Structure

ALL Projects, Areas, and Relationship notes use identical priority sections:

## High Priority / Critical
- Urgent/important items (scanned FIRST by daily planning)

## Next Actions / Current Tasks
- Regular priority items (scanned SECOND)

## Someday/Maybe
- Lower priority/exploratory (SKIPPED by daily planning)

## Waiting On
- Blocked by external dependencies

## Completed
- Finished tasks with dates

Templates

Use these templates when creating new notes:

TemplateUse When
project.mdCreating a new project (multi-step outcome)
area.mdCreating a new area of responsibility
permanent-note.mdCreating a Zettelkasten permanent note
fleeting-note.mdQuick knowledge capture for later processing
relationship.mdTracking an important person
meeting-note.mdMeeting documentation
daily-plan.mdDaily execution plan
daily-inbox.mdDaily capture file
user-context.mdUser goals and preferences
excalidraw-diagram.mdVisual diagrams and flowcharts

Key References


ADHD-Friendly Principles

The system is designed for users who may have ADHD:

  1. Read entire documents first - Understand existing structure
  2. Make targeted edits - Update specific sections, don't append
  3. Never just add to bottom - Unless explicitly asked
  4. Keep it concise - Remove redundancy
  5. One plan, not many - Replace old plans, don't add "revised" sections

Bad: Adding "REVISED PLAN" below "TODAY'S PLAN" Good: Replacing "TODAY'S PLAN" content with updated tasks


Daily Workflow Summary

The Complete Loop:

  1. Throughout day: Capture thoughts (30 sec each)
  2. 3x per week: Process inbox - Clarify + Organize (15 min)
  3. Every morning: Daily plan - Choose what to work on (5 min)
  4. Every evening: Daily closeout - Reflect + Prep tomorrow (5 min)

Total time: ~20-25 min/day + 45 min/week processing = Sustainable!


Examples

Example 1: Quick Capture

User: "Capture: Need to call the dentist and also research new project management tools"

You:

  1. Read config to get vault path
  2. Find/create today's inbox file at {{vaultPath}}/00-Inbox/Daily/YYYY-MM-DD.md
  3. Append the capture with timestamp
  4. Confirm: "Captured at 14:32. You have 3 captures today."

Example 2: Daily Planning

User: "What should I work on today?"

You:

  1. Check inbox count (process if 5+ items)
  2. Scan all projects in 01-Projects/ for next actions
  3. Ask: "How's your energy today? Any context constraints?"
  4. Generate prioritized list based on goals from user context
  5. Create/update Daily Plans/YYYY-MM-DD.md

Example 3: Proactive Capture

During conversation about a topic:

You: "That insight about [topic] seems valuable. Would you like me to capture it to your Second Brain? I can add it to your inbox for later processing, or create a permanent note if it's already well-formed."


Tool Usage

Always use the vault path from Claude Memory for all operations.

Claude Desktop (with Filesystem Extension)

Claude Desktop requires the Filesystem Desktop Extension to read/write files. Once installed, these tools are available:

Skill OperationFilesystem Extension Tool
Read filesread_file
Write new fileswrite_file
Edit existing filesedit_file
Find files by patternlist_directory + search_files
Create directoriescreate_directory

Prerequisites for Claude Desktop:

  1. Install the Filesystem extension from the built-in extension store
  2. Grant access to your Obsidian vault folder when prompted
  3. Claude will then be able to read/write files in your vault

Claude Code

Claude Code has built-in file system access. These tools are available:

  • Read - Check files, load context
  • Write - Create new files
  • Edit - Update existing files
  • Glob - Find files by pattern
  • Bash - Create directories, run commands

Capability Comparison

CapabilityClaude DesktopClaude Code
File operations in vault✅ Yes (with extension)✅ Yes (built-in)
Claude Memory✅ Yes✅ Yes
Config file fallback❌ No✅ Yes
Bash commands❌ No✅ Yes

Important: The skill is designed to work in BOTH environments:

  • Configuration is stored in Claude Memory (works everywhere)
  • File operations use the filesystem extension (Desktop) or built-in tools (Code)

Version History

Version 4.1 (2025-12-11)

  • Switched to Claude Memory for configuration persistence
  • Added Claude Desktop compatibility (sandboxed skills)
  • Legacy config file fallback for Claude Code

Version 4.0 (2025-10-15)

  • Unified skill architecture (single skill replaces multiple commands)
  • Natural language triggers instead of slash commands
  • Intelligent handling of existing Obsidian vaults
  • Works in both Claude Code and Claude Desktop

平台分布

Claude Code

30.15%
按下载量换算377

Gemini CLI

24.97%
按下载量换算312

Codex

19.32%
按下载量换算242

OpenCode

12.21%
按下载量换算153

Antigravity

6.98%
按下载量换算87

Cursor

3.87%
按下载量换算48

安全审计

Gen Agent Trust Hub

未通过

Socket

通过

Snyk

通过

权限和风险

接触敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

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。来源安全扫描存在提醒或未通过项,安装前请自行复核。

继续浏览同类 Skills