Skip to main content
Glama

Cedric is an AI employee for Slack, Telegram and Discord. He joins your meetings, works in the tools you connect (HubSpot, Notion, Gmail, Google Calendar, Linear, GitHub, Stripe and 3,000+ more) and asks before anything is sent or changed.

This repo is how you reach him from code. Cedric is a remote MCP server, so Claude Code, Cursor, VS Code, Claude, ChatGPT, Codex CLI, Gemini CLI, Windsurf or your own agent can use his 129 tools: team memory, channels, connected apps, meetings, files, workflows and scheduling. Your agent is the brain; Cedric is the hands.

Endpoint

https://app.meet-cedric.com/api/v1/mcp

Transport

Streamable HTTP

Auth

Either way works: sign in once in the browser (OAuth 2.1 with PKCE, nothing to paste), or send an API key cedric_sk_...

Account

Free to start, no card: app.meet-cedric.com

Quick start

Claude Code:

claude mcp add --transport http cedric https://app.meet-cedric.com/api/v1/mcp
# then run /mcp, pick cedric, and choose Authenticate

Cursor (~/.cursor/mcp.json):

{
  "mcpServers": {
    "cedric": { "url": "https://app.meet-cedric.com/api/v1/mcp" }
  }
}

VS Code:

code --add-mcp '{"name":"cedric","type":"http","url":"https://app.meet-cedric.com/api/v1/mcp"}'

Claude, ChatGPT, Codex CLI, Gemini CLI, Windsurf and API keys: see docs/clients.md.

Once connected, call the getting_started tool. It returns the full guide, including when to act directly and when to hand a whole task to Cedric with ask_cedric.

Related MCP server: StateMCP

What your agent can do with Cedric

A few of the 129 tools:

Area

Tools

Hand over a whole task

ask_cedric

Team memory

remember, recall_memories, search_workspace

Chat

read_channel, send_channel_message, send_dm, post_vote

Connected apps

list_connectors, integration_request, run_recipe

Email and calendar

gmail_search, gmail_send, calendar_free_busy, calendar_create_event

Meetings

join_call, share_meeting_notes, meeting_brief

Schedules and workflows

schedule_task, start_workflow, watch_app, watch_source

Files and documents

upload_file, read_file, create_document, create_presentation, create_spreadsheet

Most tools run as soon as you call them. Under a browser sign-in or a standard key, two kinds don't:

  • Direct key only: sending mail or DMs, changing the calendar, calling a connected app's API (integration_request, run_recipe), running code and compiled automations (schedule_automation) are refused with requires_direct_key. Use a direct key, or hand the task to ask_cedric, which asks a person before anything goes out.

  • Needs approval: posting in a channel, moving an event hook and starting a build return pending_approval until a person approves.

See approvals.

What people build with it

  • "What did we decide?" from the editor. Before touching the pricing code, your coding agent asks recall_memories and gets the team's actual decision, not a guess.

  • Ship and tell. When a PR merges, the agent writes the summary, send_channel_message posts it in #eng once a teammate approves, and remember keeps the decision for next time.

  • CI that speaks up. A failed deploy calls /api/v1/chat: "Tell #eng what broke and open a Linear issue." Cedric drafts both and a teammate approves with one tap.

  • Settle it with a vote. post_vote asks the channel which name to ship, and vote_results brings the tally back to your agent.

More, with the calls spelled out: docs/use-cases.md.

Use it from code

The official MCP SDKs connect with an API key from Cedric's Settings → MCP.

TypeScript (examples/node/mcp-client.mjs):

import { Client } from "@modelcontextprotocol/sdk/client/index.js";
import { StreamableHTTPClientTransport } from "@modelcontextprotocol/sdk/client/streamableHttp.js";

const transport = new StreamableHTTPClientTransport(
  new URL("https://app.meet-cedric.com/api/v1/mcp"),
  { requestInit: { headers: { Authorization: `Bearer ${process.env.CEDRIC_KEY}` } } },
);
const client = new Client({ name: "my-agent", version: "1.0.0" });
await client.connect(transport);

const { tools } = await client.listTools();
const result = await client.callTool({
  name: "recall_memories",
  arguments: { query: "pricing decisions" },
});

Python (examples/python/mcp_client.py):

from mcp import ClientSession
from mcp.client.streamable_http import create_mcp_http_client, streamable_http_client

http = create_mcp_http_client(headers={"Authorization": f"Bearer {key}"})
async with http, streamable_http_client(URL, http_client=http) as (read, write):
    async with ClientSession(read, write) as session:
        await session.initialize()
        result = await session.call_tool("recall_memories", {"query": "pricing decisions"})

No agent of your own? One HTTP call with an API key runs a whole Cedric turn (examples/node/chat.mjs, examples/python/chat.py):

curl -X POST https://app.meet-cedric.com/api/v1/chat \
  -H "Authorization: Bearer $CEDRIC_KEY" \
  -H "Content-Type: application/json" \
  -d '{"message": "Summarize this week for the team"}'

# { "reply": "...", "session": "abc123", "outcome": "answered" }

Chat sessions, file uploads, approvals and limits: docs/api.md.

In this repo

docs/clients.md          setup for every MCP client, API keys, how the sign-in works
docs/tools.md            every tool, grouped, with its parameters
docs/use-cases.md        worked examples of what to build with Cedric
docs/api.md              MCP, chat and files endpoints, approvals, limits
examples/node/           MCP SDK client and chat call (TypeScript SDK)
examples/python/         MCP SDK client and chat call (Python SDK)
examples/curl/cedric.sh  the whole API from the shell
server.json              the MCP Registry entry

Cedric is built in Zurich, Switzerland, by SFF Studio. The examples and docs in this repo are MIT licensed; the Cedric service itself is not open source.

Related MCP Connectors

Related MCP Servers

  • A
    license
    Not graded
    quality
    B
    maintenance
    MCP server for Microsoft Teams that lets AI assistants search, message, and manage chats, meetings, files, and contacts using your existing Teams browser login—no app approval or Azure AD registration needed.
    1,327 npm
    1
    MIT
  • A
    license
    Not graded
    quality
    C
    maintenance
    Enables MCP clients to discover and create agents, durably submit messages with verifiable replies, and search source-linked local memory. Exposes eight authenticated tools with owner-approved OAuth, transcript-backed verification, and explicit control over history import and embedding.
    MIT
  • A
    license
    Not graded
    quality
    B
    maintenance
    Exposes a remote Streamable HTTP MCP endpoint that lets an AI client save, search, and retrieve structured memories, archived conversations, social-account notes, and knowledge documents. Access is scoped by per-client API keys with configurable read/write permission levels, collections, and operations.
    ISC