Token导航 LogoToken导航

Agent Skill · 研究检索

obsidian-sync(Obsidian sync 工具)

obsidian-sync 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。

obsidian-sync 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,权限为调用外部服务;许可证信息还不明确,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 70,834。

总安装 198周安装 8GitHub Stars 6下载 62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
unknown

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:obsidian-sync(Obsidian sync 工具)
来源仓库:https://github.com/synapsync/synapse_registry
仓库路径:skills/obsidian-sync
安装命令:
npx skills add https://github.com/synapsync/synapse_registry --skill obsidian-sync
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 npx skills 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。

skills.shnpx skills
npx skills add https://github.com/synapsync/synapse_registry --skill obsidian-sync

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。

obsidian-sync 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。适用于多端同步、版本比对或知识库更新管理等研究检索类任务场景。通过关键词、任务描述或来源线索触发检索,返回结构化候选信息供进一步核验。

安装命令为 npx skills add https://github.com/synapsync/synapse_registry --skill obsidian-sync。使用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操作。

来源仓库内路径:skills/obsidian-sync

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/synapsync/synapse_registry --skill obsidian-sy…;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

obsidian-sync 怎么安装?

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

obsidian-sync 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。

obsidian-sync 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。

obsidian-sync 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

obsidian-sync 用的人多吗?

公开数据是累计安装 198 次,近一周安装 8 次,下载 62 次,GitHub 6 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。

obsidian-sync 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。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 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

  • 需要明确授权才能用到公司项目里的场景。
  • 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
obsidian-sync

Obsidian sync 工具 · 本页

1986调用外部服务未明确未通过

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

Purpose

Sync markdown reports, plans, and documentation generated in the workspace to an Obsidian vault using the Obsidian MCP server. Acts as the "last mile" helper that takes output produced by other skills (planners, analyzers, reporters) and persists it in the user's knowledge base.

When to Use This Skill

  • User explicitly asks to sync, save, move, or store files to Obsidian
  • User says "guardar en obsidian", "sincronizar a obsidian", or similar in Spanish
  • User asks to create a report/plan AND save it to Obsidian (this skill handles the Obsidian part after creation)
  • User wants to browse their vault to choose where to place documents
  • User asks to move an entire folder (e.g., .agents/plan/) to their Obsidian vault
  • After another skill finishes generating documents, user wants them in Obsidian

When NOT to Use

  • User wants to read or search notes already in Obsidian (use MCP tools directly)
  • User wants to edit an existing Obsidian note in-place (use MCP tools directly)
  • Files are not markdown or text-based (this skill is for .md content)

Critical Rules

RULE 1 - ALWAYS LOAD MCP TOOLS FIRST

Before calling ANY Obsidian MCP tool, you MUST use ToolSearch to load them:

ToolSearch query: "+obsidian write"    # For writing notes
ToolSearch query: "+obsidian list"     # For listing directories
ToolSearch query: "+obsidian read"     # For reading notes

Never call mcp__obsidian__* tools without loading them first via ToolSearch. They are deferred tools and will fail if not loaded.

RULE 2 - ALWAYS ASK WHERE TO SAVE

Never assume the destination folder. Always:

  1. List the vault's root directories
  2. Let the user pick or specify the destination
  3. If the user already specified a path (e.g., "save to work/my-project/plans"), confirm it exists or offer to create it

RULE 3 - PRESERVE CONTENT INTEGRITY

  • Read the source file completely before writing to Obsidian
  • Never modify the document content (headings, tables, code blocks, etc.)
  • Only ADD frontmatter metadata if the file doesn't already have it
  • If the file already has frontmatter, merge Obsidian-specific fields without overwriting existing ones

RULE 4 - GENERATE MEANINGFUL FRONTMATTER

Every note written to Obsidian MUST include frontmatter with at minimum:

---
title: "Document Title"
date: "YYYY-MM-DD"
project: "{project-name}"
type: "{document-type}"
tags: [relevant, tags, here]
---

Infer these values from:

  • title: First H1 heading in the document, or filename
  • date: File modification date, or today's date
  • project: Current working directory name or git repo name
  • type: Infer from content (e.g., "technical-debt", "strategic-analysis", "sprint-plan", "code-review", "architecture")
  • tags: Infer from project name, document type, and key topics

RULE 5 - REPORT RESULTS

After syncing, always report:

  • How many files were synced
  • The exact Obsidian paths where they were saved
  • Any files that were skipped and why

Workflow

Step 1: Identify What to Sync

Determine the source files from the user's request:

User SaysAction
"sync this report to obsidian"Identify the most recently created/discussed report in context
"move.agents/plan/ to obsidian"Glob all .md files in that directory
"save 00-strategic-analysis.md to obsidian"Sync that specific file
"sync all reports from today to obsidian"Find today's .md files in .agents/ or similar
"create plan X and save to obsidian"Wait for plan creation, then sync the output

Discovery commands:

# Find all markdown files in a directory
Glob pattern: ".agents/plan/**/*.md"

# Find recently created reports
Glob pattern: ".agents/**/*.md"

# Find specific file
Glob pattern: "**/{filename}.md"

Step 2: Load Obsidian MCP Tools

Always load the required tools before using them:

ToolSearch query: "+obsidian list"     # Loads mcp__obsidian__list_directory
ToolSearch query: "+obsidian write"    # Loads mcp__obsidian__write_note

Load both in parallel if possible. Only load read or search if you need to check existing content.

Step 3: Browse Vault and Present Options

List the vault structure and present folder options to the user:

mcp__obsidian__list_directory(path: "/")           # Root level
mcp__obsidian__list_directory(path: "/work")        # One level deeper if needed

Build a folder tree and present it using AskUserQuestion:

AskUserQuestion:
  question: "Where should I save the documents in your Obsidian vault?"
  header: "Vault folder"
  options:
    - label: "work/project-name/plans"
      description: "Inside your project's plans folder"
    - label: "work/project-name/"
      description: "At the project root in your vault"
    - label: "Create new folder"
      description: "I'll create a new folder path for you"

Important behaviors:

  • If the user already specified a destination (e.g., "save to work/agent-sync-sdk/plans"), skip the question and use that path directly
  • If a project folder already exists in the vault, suggest it as the first (recommended) option
  • Always include "Create new folder" as an option
  • If user picks "Create new folder" or "Other", ask for the desired path

Step 4: Read Source Files

For each file to sync:

  1. Read the full content using the Read tool
  2. Check if frontmatter already exists (starts with ---)
  3. Extract the first H1 heading for the title
  4. Determine the project name from cwd or git

Step 5: Write to Obsidian

For each file, call mcp__obsidian__write_note with:

mcp__obsidian__write_note(
  path: "{vault-destination}/{filename}.md",
  content: "{file-content-without-frontmatter}",
  frontmatter: {
    "title": "Document Title",
    "date": "2026-02-10",
    "project": "agent-sync-sdk",
    "type": "strategic-analysis",
    "source": ".agents/plan/2026-02-10/00-strategic-analysis.md",
    "tags": ["agent-sync-sdk", "strategy", "plan"]
  }
)

Frontmatter generation logic:

FieldSourceFallback
titleFirst # Heading in documentFilename without extension, titlecased
dateDate in filename or directory path (e.g., 2026-02-10)Today's date
projectGit repo name or directory nameWorking directory basename
typeInfer from content keywords (see table below)"note"
sourceRelative path from workspace rootAbsolute path
tagsProject name + type + key content topics[project-name]

Type inference:

Content ContainsType
"deuda tecnica", "technical debt", "code quality""technical-debt"
"estrategico", "strategic", "roadmap""strategic-analysis"
"crecimiento", "growth", "vision""growth-vision"
"sprint", "todo", "checklist""sprint-plan"
"analisis", "analysis", "report""analysis"
"arquitectura", "architecture""architecture"
"planning", "plan de trabajo""plan"
"requirements", "requisitos""requirements"
"code review", "revision""code-review"

Step 6: Report Results

After all files are written, report a summary:

Synced to Obsidian:
- work/agent-sync-sdk/plans/00-strategic-analysis.md
- work/agent-sync-sdk/plans/01-technical-debt.md
- work/agent-sync-sdk/plans/02-growth-vision.md

3 files synced successfully to vault folder: work/agent-sync-sdk/plans/

Batch Sync Pattern

When syncing an entire folder, process files efficiently:

  1. Glob all .md files in the source directory
  2. Read all files in parallel (multiple Read calls in one message)
  3. Load MCP tools once (ToolSearch)
  4. List vault directory once to build options
  5. Ask user once for destination (not per file)
  6. Write all files in sequence (Obsidian MCP may not support parallel writes safely)
  7. Report all results at the end

Integration with Other Skills

This skill is designed to be the output handler for other skills:

Producing SkillWhat It Createsobsidian-sync Role
project-planner.synapsync/planning/{project}/ plansSync planning docs to Obsidian
code-analyzer.synapsync/technical/module-analysis/ reportsSync analysis reports to Obsidian
sdlc-plannerSDLC Phase 1 & 2 documentsSync requirements & design docs
universal-plannerAdaptive planning documentsSync any planning output
Custom team analysis.agents/plan/ reportsSync team-generated reports

Composition pattern: When the user says "create a plan and save it to obsidian":

  1. The planner skill runs first and produces documents locally
  2. This skill then picks up those documents and syncs them to the vault
  3. The user only needs one instruction; the skills compose naturally

Configuration Options

By Sync Scope

ScopeCommandBehavior
Single file"sync this file to obsidian"Sync one specific file
Directory"sync.agents/plan/ to obsidian"Sync all.md files in directory
Pattern"sync all reports from today"Glob and sync matching files
Output"create X and save to obsidian"Wait for creation, then sync

By Destination Logic

ScenarioBehavior
User specifies pathUse it directly, create folders if needed
Project folder exists in vaultSuggest it as recommended option
Project folder doesn't existOffer to create work/{project-name}/
Multiple projects in vaultList all and let user choose

Best Practices

Before Sync

  • Verify the source files exist and are readable
  • Check that the Obsidian MCP server is connected (ToolSearch will fail if not)
  • If syncing output from another skill, ensure that skill has completed

During Sync

  • Never modify document content -- only add/merge frontmatter
  • Use the mcp__obsidian__write_note with mode: "overwrite" for clean writes
  • If a note already exists at the destination, warn the user before overwriting
  • Process files sequentially to avoid MCP race conditions

After Sync

  • Always report the full list of synced paths
  • If any files failed, report which ones and why
  • Suggest the user check the notes in Obsidian for visual verification

Limitations

  • Obsidian MCP required: The Obsidian REST API MCP server must be configured and running
  • Markdown only: This skill handles .md files. Binary assets, images, and non-text files are not synced
  • No bidirectional sync: This skill writes TO Obsidian, it does not pull changes back from Obsidian
  • No conflict resolution: If a note exists at the destination, it will be overwritten (with user confirmation)
  • Single vault: Operates on the vault connected to the MCP server; cannot switch between multiple vaults

Troubleshooting

IssueCauseSolution
"Tool not found: mcp__obsidian__*"MCP tools not loadedRun ToolSearch query: "+obsidian write" first
"MCP server not connected"Obsidian REST API not runningStart Obsidian and enable the REST API plugin
"Permission denied"Tool not in allowed permissionsAdd mcp__obsidian__write_note to .claude/settings.local.json
"Note already exists"Duplicate syncConfirm with user before overwriting
"Empty content"Source file doesn't existVerify path with Glob before reading
Frontmatter corruptionFile has non-standard frontmatterRead file first, detect existing frontmatter, merge carefully

Example: Full Sync Flow

User says: "sincroniza los reportes de.agents/plan/2026-02-10/ a obsidian"

Step 1 - Discover files:

Glob(".agents/plan/2026-02-10/**/*.md")
# Found: 00-strategic-analysis.md, 01-technical-debt.md, 02-growth-vision.md

Step 2 - Load MCP tools:

ToolSearch("+obsidian list")
ToolSearch("+obsidian write")

Step 3 - Browse vault:

mcp__obsidian__list_directory(path: "/")
# Found: work/
mcp__obsidian__list_directory(path: "/work")
# Found: agent-sync-sdk/, bmtz/, wdt/

Step 4 - Ask user:

AskUserQuestion:
  "Where should I save the 3 reports in your Obsidian vault?"
  options:
    - "work/agent-sync-sdk/plans (Recommended)"  # Project folder exists
    - "work/agent-sync-sdk/"                       # Project root
    - "Create new folder"

Step 5 - Read & Write:

# Read all 3 files in parallel
Read(".agents/plan/2026-02-10/00-strategic-analysis.md")
Read(".agents/plan/2026-02-10/01-technical-debt.md")
Read(".agents/plan/2026-02-10/02-growth-vision.md")

# Write each to Obsidian with generated frontmatter
mcp__obsidian__write_note(
  path: "work/agent-sync-sdk/plans/00-strategic-analysis.md",
  content: "...",
  frontmatter: { title: "Analisis Estrategico...", date: "2026-02-10", project: "agent-sync-sdk", type: "strategic-analysis", tags: [...] }
)
# ... repeat for each file

Step 6 - Report:

Synced 3 files to Obsidian:
  - work/agent-sync-sdk/plans/00-strategic-analysis.md
  - work/agent-sync-sdk/plans/01-technical-debt.md
  - work/agent-sync-sdk/plans/02-growth-vision.md

平台分布

Codex

31.79%
按下载量换算20

Claude

31.25%
按下载量换算19

Cursor

20.11%
按下载量换算12

Gemini CLI

9.38%
按下载量换算6

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

未通过

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。来源安全扫描存在提醒或未通过项,安装前请自行复核。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