mail-mcp-custom
Custom MCP server for Gmail integration, enabling email listing, reading, searching, and sending via natural language.
README
mail-mcp-custom
Custom mail MCP server for local use with opencode, without Composio. The current provider is Gmail.
Project Map
src/server.ts: MCP stdio entry point for OpenCode and MCP tool registrationsrc/web.ts: local Express server for the browser UI and local/api/*routessrc/config.ts: environment configuration, OAuth settings, token path, UI path, and Gmail scopessrc/providers/gmail/oauth.ts: local Google OAuth browser flowsrc/providers/gmail/token-store.ts: local OAuth token persistencesrc/providers/gmail/gmail.ts: Gmail API client helperssrc/providers/llm/openai-compatible.ts: local OpenAI-compatible LLM client for email summariesui/: static browser UI served bysrc/web.tsdoc/: architecture notes and OpenCode configuration
Entry Points
- MCP server for OpenCode:
npm run dev - Local browser UI:
npm run web - Type-check:
npm run check - Build:
npm run build
Documentation Maintenance
When code changes affect behavior, setup, commands, environment variables, MCP tools, web routes, security assumptions, or limitations, update the relevant docs in the same change. Check README.md, AGENTS.md, doc/architecture-diagram.md, doc/opencode-configuration.md, and .env.example before finishing.
Included Features
- Local Google OAuth for a single user
- Token persistence in
data/gmail-token.json - Basic MCP tools:
list_emailsread_emailsearch_emailssend_email
- Optional local web UI for listing, searching, reading, sending, and moving messages to Gmail Trash
- Local AI summary for emails received today through an OpenAI-compatible endpoint
Capability Matrix
| Capability | MCP Tool | Web UI/API |
|---|---|---|
| List messages | list_emails |
GET /api/emails |
| Search messages | search_emails |
GET /api/emails?query= |
| Read message | read_email |
GET /api/emails/:id |
| Send message | send_email |
POST /api/send |
| Move to trash | Not exposed | POST /api/emails/:id/trash |
| Summarize today's inbox | Not exposed | POST /api/summaries/today |
Google Cloud Setup
- Create a project in Google Cloud.
- Enable the
Gmail API. - Configure the
OAuth consent screen. - Create an
OAuth Client IDof typeWeb application. - Add this redirect URI:
http://127.0.0.1:4010/oauth/callback
Local Setup
- Copy
.env.exampleto.env. - Fill in the Google OAuth values.
- Run:
npm install
npm run dev
On the first run, your browser opens for Google authentication.
Local Web UI
Run the browser UI with:
npm run web
Then open:
http://127.0.0.1:4020
The UI serves files from ui/ and exposes local-only API routes from src/web.ts.
Optional environment settings:
WEB_HOST=127.0.0.1
WEB_PORT=4020
UI_PATH=./ui
Trash actions move messages to Gmail Trash. They do not permanently delete mail.
Local AI Summaries
The web UI can summarize emails received today with a local OpenAI-compatible LLM. The browser calls src/web.ts, and the server calls the configured local LLM endpoint.
Default local LLM settings:
LOCAL_LLM_BASE_URL=http://127.0.0.1:1234/v1
LOCAL_LLM_API_KEY=local
LOCAL_LLM_MODEL=qwen/qwen3.5-9b
LOCAL_LLM_TIMEOUT_MS=180000
LOCAL_LLM_MAX_TOKENS=4096
The summary flow searches Gmail for today's inbox messages, reads each message, truncates long bodies, sends the summary prompt to the local LLM, and filters the visible message list to the summarized messages.
Initial Scopes
gmail.readonlygmail.sendgmail.modify
gmail.modify is required for moving messages to Trash from the local web UI.
OpenCode Config
The OpenCode configuration for this project is documented separately in doc/opencode-configuration.md.
Current Limitations
- Single local user only
- Plain text email sending only
- No attachments
- Trash support only moves messages to Gmail Trash; permanent delete is not implemented
- AI summaries currently support only the
local_llmprovider; OpenCode provider switching is planned but not wired yet - No custom refresh-token logic outside the Google client
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.