FlareMo MCP
Serverless notes on Cloudflare Workers with an MCP endpoint and a skill so agents can search and write your notes
- 346 ★ on the repository
- Last commit
- AGPL-3.0
Broad access
Why this level
- A memos_pat_ token lets the agent read, create, edit and delete all your notes
- Notes often hold personal records, and the agent forwards their content to the model
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add mcp/flaremoDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- A memos_pat_ token lets the agent read, create, edit and delete all your notes
- Notes often hold personal records, and the agent forwards their content to the model
This is third-party code. Review the repository files before installing.
What it does
FlareMo deploys into your own Cloudflare account as a serverless second brain: code and static assets run on Workers, and notes, tasks and articles live in D1 and R2. It has two parts for agents. The /mcp endpoint exposes list, search, read, create, update and delete operations over notes, with access granted by a revocable token like memos_pat_. The flaremo-notes skill in the skills folder teaches an agent to search before writing, keep your wording and return the id of the note it touched. This is the only source of knowledge: the AI has no separate memory and nothing is extracted from conversations in the background. An agent reads notes on demand and writes to them only when you ask or when a result needs saving. Deletion is soft by default, and a hard delete is allowed only on an explicit request. The product is wider than MCP: card notes with tags, tasks and projects with kanban boards, publishable articles, team mode, import from Memos and full export.
Who it is for. People who keep notes and tasks in their own deployment and want Claude Code, Cursor or Codex to find context there and save outcomes between sessions.
Good fit when
- An agent should recall what you decided about a migration two weeks ago and find it by keywords in your notes
- A session's outcome should be saved as a tagged note and its id returned for checking
- Your notes live in Memos and you are moving them to a place with no separate server
- You want a shared team knowledge base with roles and three visibility levels
- You want the agent to reach only your notes, with no memory store of its own
Not a fit when
- You have no Cloudflare account and no wish to deploy a Worker, D1 and R2: the product installs only there
- Notes must not leave your own hardware: FlareMo keeps data in Cloudflare infrastructure
- You want an automatic conversation log in agent memory: the skill explicitly forbids background filling
- A third-party client does not support remote MCP over HTTP with an Authorization header
Example request
Find the FlareMo notes about the database migration, give their ids and append a summary of our conversation to the latest oneMCP
- Transport
- http
- Authentication
- API key
Environment variables
FLAREMO_URLrequired- Address of your FlareMo instance, for example https://flaremo.example.com
FLAREMO_PATrequired, secret- Access token like memos_pat_, created in account settings and shown once
Limitations
- The endpoint address depends on your instance: ${FLAREMO_URL}/mcp. The agent docs name no shared public address.
- The memos_pat_ token is created in account settings and shown once. A 401 or 403 means an expired token or an Origin outside the allow list.
- The server uses a JSON subset of Streamable HTTP with no MCP sessions. Older clients use the /api/v1/mcp path.
- The license is AGPL-3.0: running a modified version over a network extends obligations to its code.
How to disable. Revoke the token in the FlareMo account settings and the agent loses access. Delete the flaremo entry from the client's mcpServers and remove the skill from the skills directory. Delete the Worker in Cloudflare to stop the deployment.
FAQ
How do I connect FlareMo to Claude Code or Cursor?
Create a memos_pat_ token in account settings and add an http MCP server to the client config at your instance address with the /mcp path and an Authorization: Bearer header. Copy the flaremo-notes skill from the skills folder into the agent's skills directory.
Which tools does the FlareMo MCP server provide?
List and search notes with memo_list_memos, read with memo_get_memo, create with memo_create_memo, update with memo_update_memo, delete with memo_delete_memo, attachments with memo_list_memo_attachments, comments with memo_list_memo_comments and an identity check with auth_get_current_user.
Does the AI keep its own memory?
No. Your notes remain the only source of knowledge, and the agent reads them on request. Nothing is extracted from conversations in the background, and context from past sessions has to be found by keywords or tags.
How do I deploy FlareMo without my own server?
Press Deploy to Cloudflare in the README and the repository is cloned into your GitHub while D1, R2, Queues and Vectorize are created automatically. Other paths exist: a GitHub Action in a fork, deployment by an agent following docs/en/agent-deploy.md and three manual steps with wrangler.
Can I connect FlareMo to ChatGPT and other clients?
Yes, the README names Claude, Cursor, Codex, ChatGPT and ZCode. Older clients can use the /api/v1/mcp path with the tool names list_memos, create_memo, get_memo and search_memos.
Discussion
No questions or reviews yet.
Related
MCP serversYour own memory for AI tools on Cloudflare: save once, recall by meaning in Claude, ChatGPT, Cursor and any MCP client
A self-hosted scheduler on Cloudflare Workers with a built-in MCP server: an agent lists slots, creates and reschedules bookings
An agent searches, reads and writes a Roam Research graph through the desktop app's local API, with a CLI and a syntax skill
A local assistant with an MCP bridge: Claude Code and Cursor search its memory, create tasks, send notifications and ask it questions
The official Kibela MCP server searches, reads and edits notes, folders and comments of your wiki straight from an assistant chat
A local web panel where your agent builds widgets and screens on top of project files, APIs and MCP servers
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| FlareMo (this card) | yes | API key | 346 | |
| Pinto Notes | yes | OAuth | 30 | notes for iPhone and Mac, not a deployment in your own Cloudflare |
| Open Second Brain | yes | not required | 511 | agent memory in Obsidian-style Markdown files, with no web app or team mode |
README in short
The README promises a serverless second brain and compares it with home and cloud hosting. It then describes features: quick capture, MCP and a skill, tasks, articles, team mode, offline use, security and plugins. Quick deployment paths come before the stack and the AGPL-3.0 license. The README is translated into eight languages, Russian included.