What Gmail API permissions does this server require?
It uses the OAuth scope `https://www.googleapis.com/auth/gmail.modify`, which allows reading, sending, trashing, and marking emails.
@jasonsum
Basic information
Model Context Protocol (MCP) server for Gmail
Add this server to your MCP-compatible client using the configuration below.
{
"mcpServers": {
"gmail-mcp-server-jasonsum": {
"command": "npx",
"args": [
"@modelcontextprotocol/inspector",
"uv",
"run",
"[absolute-path-to-git-repo]/src/gmail/server.py",
"--creds-file-path",
"[absolute-path-to-credentials-file]",
"--token-path",
"[absolute-path-to-access-tokens-file]"
]
}
}
}Sends email to email address recipient
Moves email to trash
Marks email as read
Retrieves unread emails
Retrieves given email content
Open email in browser
This MCP server integrates with Gmail to enable sending, removing, reading, drafting, and responding to emails. It requires user confirmation before executing any email actions.
Set up a Google Cloud project, enable the Gmail API, configure OAuth consent with the gmail.modify scope, and download OAuth credentials. Pass the credentials file path and a token storage path as --creds-file-path and --token-path when starting the server. Use uv to run with Claude Desktop or MCP Inspector.
It uses the OAuth scope `https://www.googleapis.com/auth/gmail.modify`, which allows reading, sending, trashing, and marking emails.
On first run, the server launches a browser for OAuth consent. Tokens are then saved to the path specified by `--token-path` and reused on subsequent starts.
Yes. Add an `mcpServers` entry in your Claude Desktop config JSON. The command uses `uv` with parameters `--directory`, `--creds-file-path`, and `--token-path`.
Use the MCP Inspector from the repository root: `npx @modelcontextprotocol/inspector uv run <path>/server.py --creds-file-path <path> --token-path <path>`.
Both are stored locally in files you specify. The credentials file is the OAuth client JSON from Google Cloud; the token file is generated after authentication.

Messaging rooms for AI agents: hand off context across tools, worktrees, machines, and teammates. Agents meet in shared rooms over MCP, hand off directly, and leave a replayable transcript; live sessions wake on reply, b

Official Mobile Text Alerts server. Send and schedule SMS, manage subscribers and groups, and complete carrier registration from any MCP-compatible AI assistant.

Draft, schedule, and publish social media posts to 8 platforms from any AI agent: Facebook, Instagram, X, TikTok, YouTube, LinkedIn, Threads, and Bluesky. 19 tools covering drafting, scheduling, publishing, media upload,

LinkedIn MCP server for Claude, ChatGPT, Cursor & Codex. Read LinkedIn DMs, search your network, enrich profiles and draft messages you review before sending.

Caulo is the social layer of the agentic web: a social network your AI agent reads, briefs you on, and posts to — by prompt. Ad-free, verified humans only, with provenance on every post (human / agent-assisted / agent-au
Comments