Skip to main content

MCP integration

mokumokuren embeds an MCP (Model Context Protocol) server. Connect an MCP client — Claude Desktop / Codex / OpenCode — and the AI can send its drafts straight to the Inbox via the submit_drafts tool.

mokumokuren doesn't care which LLM you use. The AI client generates; mokumokuren receives and delivers.

:::info Pro feature MCP integration is a Pro feature. Make sure your account shows Pro. → Pro :::

Check that Pro is active

Enabling

  1. Open ⚙ (Settings) in the sidebar.
  2. Turn on "Enable MCP integration" (requires Pro). The status switches to "Running". Enabling MCP integration
  3. The connection URL, token, and connection snippets appear. Connection details The server listens on 127.0.0.1 (local only) and requires a local token. It runs only while the app is open.

Connecting clients

Copy a connection snippet from the settings screen and paste it into your client's configuration. There are two kinds:

  • HTTP-capable clients — connect directly with the URL and token.
  • stdio clients (Codex, etc.) — use the bundled bridge. command points at the app's executable and env carries the URL and token (no separate node install needed).

Connection snippets

Where to paste, for example:

  • Claude Desktop — mcpServers in ~/Library/Application Support/Claude/claude_desktop_config.json
  • Codex — Codex's MCP settings

Or simply click the write-to-config button for your AI client, then restart the client.

Connection buttons

After configuring, restart the client (it fetches the tool list on connect). Once connected you'll see submit_drafts / list_accounts / list_drafts.

Using it

Ask your AI client to "write post drafts and send them to mokumokuren" — the client calls submit_drafts and the drafts appear in the Inbox.

From writing to receiving

To skip saying it every time, add a standing rule to the client (e.g. Codex's AGENTS.md):

Whenever you write (or are asked to write) social-media post drafts, always send them to
mokumokuren with submit_drafts after generating.
Each draft: text (required) / title / format (text|markdown) / serviceHints.
If target services are specified, set them in serviceHints.

Token & port

  • Regenerate token — from the settings screen (e.g. if it leaks). After regenerating, update your clients' configs too.
  • Port — automatic by default; falls back to a free port when occupied.

What happens to submitted drafts

Drafts that land in the Inbox are routed and delivered per Inbox: ingestion & delivery.