Quick Start

Connect via MCP

SynquoRum implements the Model Context Protocol (MCP) on /api/mcp/streamable, so any MCP-capable agent CLI or IDE can read your projects, write notes / minutes, run integrations, and route prompts through your BYOK providers — all over a single Bearer token you control from Settings.

1. Get a token, 2. Paste config, 3. Done

Open Settings → Agent CLI, click Generate new token, pick your client (Claude Code, Cursor, Gemini, …). The page shows a token plus a ready-to-paste config block for that exact client. Copy both into your CLI's config file. Restart the CLI. Done.

Tokens are shown once — save them before closing the dialog. Revoke any token from the same screen at any time.

Supported clients

What you get

22 tools across read, write, and AI / integration. Every connected MCP client sees the same surface — the differentiator is how the client chooses to surface them (auto-discovery, slash command, tool picker, …).

* v7.0: call_ai routes a prompt through one of your BYOK providers (32 formats supported). execute_integration runs any of your 664 connected integration actions (Slack, Notion, GitHub, Gmail, …).

🟠 Claude Code

Anthropic's terminal coding agent. Native MCP HTTP transport.

Setup steps

  1. Install Claude Code (npm i -g @anthropic-ai/claude-code, or via the Claude.ai download page).
  2. Generate a token from Settings → Agent CLI, agent type = Claude Code.
  3. Paste the config block into ~/.claude.json (or %USERPROFILE%\.claude.json on Windows).
  4. Run `claude` in any terminal. Ask: 'Use the synquorum tools to list my projects'.

Config example

Generated dynamically with your token inserted at Settings → Agent CLI. The static example below is for reference; copy the live one from Settings to avoid typos.

~/.claude.json
{
  "mcpServers": {
    "synquorum": {
      "type": "http",
      "url": "https://synquorum.com/api/mcp/streamable",
      "headers": {
        "Authorization": "Bearer syn_XXXX_PASTE_YOURS_HERE"
      }
    }
  }
}

Troubleshoot

🟧 Claude Desktop

Anthropic's desktop app. Add the MCP server through the JSON config file; the desktop UI then surfaces tools in the 🔧 picker.

Setup steps

  1. Find your config: macOS: ~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json · Linux: ~/.config/Claude/claude_desktop_config.json.
  2. Generate a token from Settings → Agent CLI, agent type = Claude Desktop.
  3. Paste the config block. If the file doesn't exist, create it with the JSON below.
  4. Fully quit and reopen Claude Desktop. The 🔧 icon shows the SynquoRum tools.

Config example

Generated dynamically with your token inserted at Settings → Agent CLI. The static example below is for reference; copy the live one from Settings to avoid typos.

claude_desktop_config.json
{
  "mcpServers": {
    "synquorum": {
      "type": "http",
      "url": "https://synquorum.com/api/mcp/streamable",
      "headers": {
        "Authorization": "Bearer syn_XXXX_PASTE_YOURS_HERE"
      }
    }
  }
}

Troubleshoot

🟢 OpenAI Codex CLI

OpenAI's terminal agent. Register the MCP server via `codex mcp add` (CLI) or the config TOML.

Setup steps

  1. Install Codex CLI (`npm i -g @openai/codex` or via the OpenAI Platform download).
  2. Generate a token from Settings → Agent CLI, agent type = Codex CLI.
  3. Run the `codex mcp add` command shown in the config block, or paste the TOML into ~/.codex/config.toml.
  4. In any Codex session: `/mcp synquorum get_context`.

Config example

Generated dynamically with your token inserted at Settings → Agent CLI. The static example below is for reference; copy the live one from Settings to avoid typos.

~/.codex/config.toml
codex mcp add \
  --transport http \
  --header "Authorization: Bearer syn_XXXX_PASTE_YOURS_HERE" \
  synquorum https://synquorum.com/api/mcp/streamable

# Or add manually to ~/.codex/config.toml:
[[mcp_servers]]
name = "synquorum"
transport = "http"
url = "https://synquorum.com/api/mcp/streamable"
headers = { Authorization = "Bearer syn_XXXX_PASTE_YOURS_HERE" }

🔵 Gemini CLI

Google's terminal agent. Native MCP HTTP transport via settings.json.

Setup steps

  1. Install Gemini CLI (`npm i -g @google/gemini-cli` or via Google AI Studio).
  2. Generate a token from Settings → Agent CLI, agent type = Gemini CLI.
  3. Paste the config block into ~/.gemini/settings.json (create if missing).
  4. In any Gemini session: 'List my SynquoRum projects using the synquorum MCP server'.

Config example

Generated dynamically with your token inserted at Settings → Agent CLI. The static example below is for reference; copy the live one from Settings to avoid typos.

~/.gemini/settings.json
{
  "mcpServers": {
    "synquorum": {
      "httpUrl": "https://synquorum.com/api/mcp/streamable",
      "headers": {
        "Authorization": "Bearer syn_XXXX_PASTE_YOURS_HERE"
      }
    }
  }
}

⚡ Cursor

AI-first code editor with MCP support in Composer.

Setup steps

  1. Open Cursor → Settings → Features → MCP.
  2. Either click 'Add MCP server' in the UI, or paste the JSON into ~/.cursor/mcp.json (global) or .cursor/mcp.json (project).
  3. Generate a token from Settings → Agent CLI, agent type = Cursor.
  4. Restart Cursor. The server should appear with a green dot. In any Composer chat: '@synquorum get_context'.

Config example

Generated dynamically with your token inserted at Settings → Agent CLI. The static example below is for reference; copy the live one from Settings to avoid typos.

~/.cursor/mcp.json
{
  "mcpServers": {
    "synquorum": {
      "url": "https://synquorum.com/api/mcp/streamable",
      "headers": {
        "Authorization": "Bearer syn_XXXX_PASTE_YOURS_HERE"
      }
    }
  }
}

Troubleshoot

♾️ Continue.dev

Open-source AI coding extension for VS Code and JetBrains. MCP is under experimental.

Setup steps

  1. Install Continue from the VS Code marketplace (or JetBrains plugin store).
  2. Generate a token from Settings → Agent CLI, agent type = Continue.dev.
  3. Add the experimental.modelContextProtocolServers block to ~/.continue/config.json.
  4. Reload the Continue extension (Cmd/Ctrl+Shift+P → 'Continue: Reload Window'). Tools appear in the @ picker.

Config example

Generated dynamically with your token inserted at Settings → Agent CLI. The static example below is for reference; copy the live one from Settings to avoid typos.

~/.continue/config.json
{
  "experimental": {
    "modelContextProtocolServers": [
      {
        "transport": {
          "type": "http",
          "url": "https://synquorum.com/api/mcp/streamable",
          "headers": {
            "Authorization": "Bearer syn_XXXX_PASTE_YOURS_HERE"
          }
        }
      }
    ]
  }
}

🌊 Windsurf (Codeium)

Codeium's agentic IDE. Cascade workflows can call MCP tools automatically.

Setup steps

  1. Open Windsurf → Settings → Cascade → MCP servers.
  2. Generate a token from Settings → Agent CLI, agent type = Windsurf.
  3. Paste the config block into ~/.codeium/windsurf/mcp_config.json.
  4. Reload Windsurf. The synquorum entry should show as connected. In any Cascade chat, describe the task — Windsurf decides when to call SynquoRum tools.

Config example

Generated dynamically with your token inserted at Settings → Agent CLI. The static example below is for reference; copy the live one from Settings to avoid typos.

~/.codeium/windsurf/mcp_config.json
{
  "mcpServers": {
    "synquorum": {
      "serverUrl": "https://synquorum.com/api/mcp/streamable",
      "headers": {
        "Authorization": "Bearer syn_XXXX_PASTE_YOURS_HERE"
      }
    }
  }
}

🤝 Aider

Pair-programming terminal agent that edits your codebase via git diffs. MCP support is evolving; works best in /architect mode.

Setup steps

  1. Install Aider (`pip install aider-chat` or via uv).
  2. Generate a token from Settings → Agent CLI, agent type = Aider.
  3. Either pass --mcp-server / --mcp-header flags, or add them to ~/.aider.conf.yml.
  4. Start aider in /architect mode for multi-file edits driven by SynquoRum context.

Config example

Generated dynamically with your token inserted at Settings → Agent CLI. The static example below is for reference; copy the live one from Settings to avoid typos.

~/.aider.conf.yml
aider --mcp-server "synquorum=https://synquorum.com/api/mcp/streamable" \
      --mcp-header "synquorum:Authorization: Bearer syn_XXXX_PASTE_YOURS_HERE"

# Or via ~/.aider.conf.yml:
mcp-server:
  - synquorum=https://synquorum.com/api/mcp/streamable
mcp-header:
  - "synquorum:Authorization: Bearer syn_XXXX_PASTE_YOURS_HERE"

⚙️ Other / custom MCP client

Anything that speaks Streamable HTTP MCP (Zed, custom scripts, future tools). The endpoint contract is the same.

Setup steps

  1. Endpoint: https://synquorum.com/api/mcp/streamable
  2. Transport: Streamable HTTP (POST / GET / DELETE per the MCP spec).
  3. Auth header: Authorization: Bearer <your token>.
  4. Generate the token from Settings → Agent CLI, agent type = Other.

Config example

Generated dynamically with your token inserted at Settings → Agent CLI. The static example below is for reference; copy the live one from Settings to avoid typos.

custom
# Pseudo-config — adapt to your client's format
endpoint: https://synquorum.com/api/mcp/streamable
transport: streamable_http
auth:
  type: bearer
  token: syn_XXXX_PASTE_YOURS_HERE

Common questions

Where does call_ai get its API key?

From your registered providers in Settings → AI Providers. Pick a provider UUID; the server decrypts your BYOK key and calls the upstream model. The MCP token alone is not a model-provider key — it's a SynquoRum-session key that authorizes use of your stored credentials.

Does call_ai consume SynquoRum points?

β1.0: no — MCP calls bypass the chat UI's point pipeline. Cost is borne by the upstream provider on your BYOK key. β1.1+ will unify this so MCP and chat-UI calls share the same metering.

Are integration calls audited?

β1.0: chat-UI calls write to integration_logs; MCP calls currently don't. The library refactor that landed in ラリー 189-190 sets up the shared call path; the audit-log unification ships in β1.1+.

Rate limits?

120 MCP requests per minute per token. The 30/min limit on chat-UI /api/ai/send does not apply to the MCP path — they're separate accounting buckets.

Lost the token — can I retrieve it?

No. Tokens are hashed on creation and the plaintext is never stored. Revoke the lost token from Settings → Agent CLI and generate a new one. The revoked token stops working immediately.

MCP Quick Start | SynquoRum | SynquoRum