@withrigor/mcp-server

Model Context Protocol server for Rigor — lets users summon their coding agent from inside a Rigor project without switching to their IDE.

This is the transport layer that connects your coding agent (Claude Code, Codex, Cursor, any MCP-compatible client) to a Rigor project. Once connected, the agent can:

• Read the full project context in one call (rigor_get_pickle)
• Claim pending tasks (/rigor runs, visualize requests, comment replies)
• Write markdown docs with rich blocks (charts, metrics, mermaid, tables)
• Query the connected database
• Post annotations, suggestions, and replies to comment threads

Previously, an external agent had to be invoked manually via a terminal command (/rigor). With MCP, the agent is connected once and watches for work inside the session — users click a "Summon" button in Rigor and the agent picks up the task immediately.

Install

Via Claude Code / Claude Desktop

Add to your ~/.claude/claude_desktop_config.json (Claude Desktop) or .mcp.json in your project root (Claude Code):

{
  "mcpServers": {
    "rigor": {
      "command": "npx",
      "args": ["-y", "@withrigor/mcp-server"],
      "env": {
        "RIGOR_BASE_URL": "https://withrigor.dev",
        "RIGOR_API_TOKEN": "rg_...",
        "RIGOR_AGENT_ID": "claude-code"
      }
    }
  }
}

Get your RIGOR_API_TOKEN from the /agent-setup page in your Rigor project.

Local development

cd mcp
pnpm install
pnpm build

Then point your MCP client at node /absolute/path/to/mcp/dist/server.js.

Environment variables

| Variable | Required | Default | Description | |---|---|---|---| | RIGOR_BASE_URL | no | https://withrigor.dev | Base URL of the Rigor instance. Use http://localhost:3000 for local dev. | | RIGOR_API_TOKEN | yes | — | Per-project bearer token. From /agent-setup. | | RIGOR_AGENT_ID | no | mcp:coding-agent | Identifier for this agent in audit logs and presence indicators. |

Tools exposed

| Tool | Purpose | |---|---| | rigor_get_pickle | One-shot project context: schema, formulas, themes, feature set | | rigor_get_state | Full project state including doc bodies | | rigor_get_pending | Lightweight list of pending work | | rigor_watch_tasks | Long-poll (up to 120s) for new tasks | | rigor_claim_task | Atomically claim a task before working on it | | rigor_submit_task_result | Return a structured result for a claimed task | | rigor_write_doc | Create or update a plan document (markdown) | | rigor_annotate | Post an intervention chip on a doc | | rigor_reply | Reply to an existing comment thread | | rigor_query_db | Read-only SELECT against the connected DB | | rigor_render_mermaid | Render mermaid to SVG |

Typical session loop

1. rigor_get_pickle({ project_id }) → baseline context
2. rigor_get_pending() → any work waiting?
3. For each pending task:
   a. rigor_claim_task({ notification_id })
   b. Execute (write doc / build chart / answer comment / …)
   c. rigor_submit_task_result({ notification_id, result_type, result_data })
4. rigor_watch_tasks({ project_id }) → block for new work
5. Loop.

Design notes

• stdio transport — matches Claude Desktop, Claude Code, Cursor conventions
• Per-project scope — one token = one project. Run multiple servers for multiple projects.
• Long-poll client-side — Rigor's HTTP API isn't long-poll native, so rigor_watch_tasks polls /api/coplanner/agent/pending every 2s up to the deadline. Simple and reliable.
• SQL guard — rigor_query_db rejects anything that doesn't start with WITH or SELECT. Write/DDL queries are never allowed through the MCP.