note-mcp
Unofficial MCP server for note.com using cookie-based authentication to manage notes and drafts via internal APIs.
README
note-mcp-community
Unofficial, community-maintained stdio MCP server for note.com. It uses cookie-based access to note.com's internal APIs.
[!WARNING] This project is unofficial and not affiliated with note.com. Internal APIs can change without notice. Keep cookies local and never commit them to GitHub, npm, logs, or issue reports.
Quick start
Local/desktop agents
Use browser login first:
npx note-mcp-community auth
After you log in to note.com in the opened browser, note-mcp-community saves the cookie locally and prints the saved file path:
{
"authenticated": true,
"saved": true,
"configPath": "/Users/you/.config/note-mcp-community/config.json",
"cookiePreview": "fp=b…5948",
"message": "note.com authentication configured from browser login. Cookie saved to /Users/you/.config/note-mcp-community/config.json."
}
Then configure your MCP client without putting cookies in the config:
{
"mcpServers": {
"note": {
"command": "npx",
"args": ["-y", "note-mcp-community"]
}
}
}
Some MCP clients only load newly added tools when a process or conversation starts. After authentication or config changes, restart the client or open a new thread/session if note_* tools do not appear immediately.
Quick setup check:
npx note-mcp-community authnpx note-mcp-community auth --status- Add
npx -y note-mcp-communityto your MCP client config - Restart the client or open a new thread/session
- Run
note_auth_statusfrom the MCP client
Servers, containers, and CI
Do not rely on browser login in headless/container environments. Provide a cookie through env or a mounted config file instead:
NOTE_COOKIE='your note.com Cookie header' npx note-mcp-community
Or mount a config file and point NOTE_MCP_COMMUNITY_CONFIG at it:
docker run \
-v ~/.config/note-mcp-community/config.json:/run/secrets/note-mcp-community-config.json:ro \
-e NOTE_MCP_COMMUNITY_CONFIG=/run/secrets/note-mcp-community-config.json \
your-agent-image
Install / run
npx note-mcp-community
For local development:
npm install
npm run build
node dist/index.js
MCP client configuration
Recommended desktop setup after npx note-mcp-community auth:
{
"mcpServers": {
"note": {
"command": "npx",
"args": ["-y", "note-mcp-community"]
}
}
}
Advanced env-based setup:
{
"mcpServers": {
"note": {
"command": "npx",
"args": ["-y", "note-mcp-community"],
"env": {
"NOTE_COOKIE": "your note.com Cookie header"
}
}
}
}
Codex CLI
For Codex, add note-mcp-community as a stdio MCP server with npx:
codex mcp add note -- npx -y note-mcp-community
codex mcp list
npx -y note-mcp-community auth --status
Recommended verification flow:
npx note-mcp-community authnpx note-mcp-community auth --statuscodex mcp add note -- npx -y note-mcp-community- Restart Codex or create a new thread
- Run
note_auth_status
If the server is listed but note_* tools are not available in the current thread, restart Codex or start a new thread so the MCP tools are loaded from the new configuration.
Authentication
note-mcp-community supports two authentication paths.
1. Local/desktop: browser login
For local desktop agents, ask the agent to call:
note_auth_login
Or run it directly:
npx note-mcp-community auth
This opens a browser, lets you log in to note.com normally, then stores note.com cookies in:
~/.config/note-mcp-community/config.json
The CLI and MCP tool response include the actual configPath used.
If the browser executable is not installed yet, install Playwright's Chromium once on the same machine/user account, then retry:
npx playwright install chromium
When using note-mcp-community only through npx and Playwright is not otherwise installed globally/in the project, this form is often more reliable:
npx -p playwright playwright install chromium
For remote servers, containers, or CI, prefer the secret/env/config-file path below instead of browser login.
Useful CLI commands:
npx note-mcp-community auth --status
npx note-mcp-community auth --clear
npx note-mcp-community auth --headless
npx note-mcp-community auth --headed
2. Advanced/server/CI: secret, env, or config file
For remote agents, servers, containers, CI, and secret managers, provide a Cookie header via:
NOTE_COOKIENOTE_SESSION_COOKIENOTE_MCP_COMMUNITY_CONFIGpointing to a config JSON file- MCP tool
note_set_cookie
Example config file:
{
"cookie": "your note.com Cookie header",
"updatedAt": "2026-06-21T00:00:00.000Z"
}
Cookie lookup priority:
NOTE_COOKIENOTE_SESSION_COOKIE- config file cookie
Default config path:
~/.config/note-mcp-community/config.json
Override config path:
NOTE_MCP_COMMUNITY_CONFIG=/path/to/config.json npx note-mcp-community
Tools
Authentication/setup tools:
note_auth_status— inspect whether auth is configured and which config path is usednote_auth_login— open a browser login flow and save cookies locally; response includesconfigPathnote_set_cookie— save a Cookie header to the local config file, optionally verifying it firstnote_clear_cookie— delete the stored config-file cookienote_login_help— explain supported setup paths
note.com tools:
note_auth_check— verify configured cookie-based access to note.com internal APIsnote_list_my_notes— list notes for the authenticated account viaGET /v2/note_list/contents?limit={limit}&page={page}. By default returns the full internal API payload. For LLM-friendly list views, passfields: "summary"orincludeBody: falseto return summary fields such astitle,key,url,publishAt,status,likeCount, andisAuthor.note_list_drafts— list drafts for the authenticated account viaGET /v2/note_list/contents?limit={limit}&page={page}&status=draft&without_magazines=true. By default returns the full internal API payload. For LLM-friendly list views, passfields: "summary"orincludeBody: false.note_get_note— fetch a note by note key, e.g.n1a0b26f944f4note_get_draft— fetch authenticated draft detail by note key viaGET /v3/notes/{noteKey}?draft=true&draft_reedit=falsenote_create_draft— create a draft by first callingPOST /v1/text_noteswith an empty-body editor payload to obtaindata.id/data.key, then saving content viadraft_save. By default returns an LLM-friendly summary withid/noteId,key/noteKey,editUrl,publicUrl, andnextActions; passresponseFormat: "full"for raw responses.note_update_draft— update a draft by numeric draft/note id. By default returns a compact summary; passresponseFormat: "full"for raw responses.note_publish_draft— publicly publish a draft by note key; internally resolves the numeric id from draft detail, then callsPUT /v1/text_notes/{id}with note.com's current publish payload. By default returnsstatus,key,noteUrl,eyecatch, andpublishedAt; passresponseFormat: "full"for raw responses.note_upload_eyecatch— upload an eyecatch/cover image viaPOST /v1/image_upload/note_eyecatch. Use thedraft.idreturned bynote_create_draftasnoteId; this can be called before publishing. Provide numericnoteIdand eitherimagePathorimageUrl; width/height default to note.com's recommended1280x670. By default returnsnoteIdandeyecatchUrl; passresponseFormat: "full"for raw responses.note_delete_draft— delete an unpublished draft by numeric draft/note id viaDELETE /v1/text_notes/draft_delete?id={draftId}note_delete_note— delete a published/deletable note by note key viaDELETE /v1/notes/n/{noteKey}
If authentication is missing, note tools return an auth_required error suggesting note_auth_login or note_set_cookie.
Body format for AI agents
note_create_draft and note_update_draft send body directly to note.com's internal editor API. note.com does not automatically render Markdown in this field.
If you want headings, lists, links, or emphasis to appear formatted, pass note-compatible HTML:
<h2>テスト内容</h2>
<ul>
<li>下書き作成</li>
<li>公開</li>
</ul>
Do not pass Markdown if visual formatting is expected:
## テスト内容
- 下書き作成
- 公開
AI agents should generate or convert content to note-compatible HTML before calling the tool. note-mcp-community intentionally stays a thin bridge to note.com's API; Markdown-to-HTML conversion belongs in the caller or a future optional helper, not in the core draft tools.
Recommended body HTML:
<h2>,<h3>for headings<p>for paragraphs<ul><li>/<ol><li>for lists<strong>,<em>for emphasis<a href="...">for links
Avoid:
- Full HTML documents (
<html>,<head>,<body>) - Inline scripts/styles
- Unsupported custom attributes
API basis
The initial endpoints are based on public, unofficial note API references, including:
Known endpoint basis:
- Base URL:
https://note.com/api - Note detail:
GET /v3/notes/{noteKey} - Draft detail:
GET /v3/notes/{noteKey}?draft=true&draft_reedit=false&ts={timestamp} - Authenticated note list:
GET /v2/note_list/contents?limit=20&page=1 - Authenticated draft list:
GET /v2/note_list/contents?limit=20&page=1&status=draft&without_magazines=true - Draft shell create/id lookup:
POST /v1/text_noteswith{ "body": "", "body_length": 0, "name": "...", "index": false, "is_lead_form": false }; response includes numericdata.idand notedata.key. Mutating editor requests should includeOrigin: https://editor.note.com,Referer: https://editor.note.com/,X-Requested-With: XMLHttpRequest, andContent-Type: application/json. - Draft save/update:
POST /v1/text_notes/draft_save?id={draftId}&is_temp_saved=truewithbody,body_length,name,index, andis_lead_form - Draft publish:
PUT /v1/text_notes/{draftId}withfree_body,pay_body,body_length, andstatus: "published" - Eyecatch upload:
POST /v1/image_upload/note_eyecatchas multipart/form-data withnote_id, binaryfile,width, andheight. The default/recommended image dimensions are1280x670. - Draft delete:
DELETE /v1/text_notes/draft_delete?id={draftId} - Published/deletable note delete:
DELETE /v1/notes/n/{noteKey} - Auth smoke test:
GET /v3/notice_counts
note_list_my_notes and note_list_drafts intentionally expose the authenticated note list endpoints above. The response shape is determined by note.com's internal API and typically returns items under data.notes; use the default full response when debugging endpoint behavior, and use summary mode when a compact list is enough. Summary mode does not invent public URLs for drafts unless note.com returns an explicit URL/path.
Release
Releases are handled by GitHub Actions + semantic-release.
- Push or merge Conventional Commits into
main. - GitHub Actions runs CI.
- The release workflow creates GitHub tags/releases and publishes to npm.
npm publishing uses npm Trusted Publishing with GitHub Actions OIDC. Configure new-village/note-mcp-community and .github/workflows/release.yml as a trusted publisher on npmjs.com. No NPM_TOKEN repository secret is required.
Development
npm run typecheck
npm test
npm run build
npm run lint
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.