The Model Context Protocol (MCP) is becoming the standard way AI agents talk to the tools you already use. Risal ships a first-class MCP server, so the same agent you code or chat with can read and change your tasks, sprints, timers and servers — no glue scripts, no brittle browser automation.
What it unlocks
Once connected, you can ask your agent in plain language:
- “Create task NOT-42 in the mobile project and start a sprint.”
- “What’s still open in the current sprint, and who’s it assigned to?”
- “Log two hours against the API refactor.”
- “Restart the Docker stack on the staging server and tell me the disk usage.”
The agent calls the Risal MCP tools directly. Because the protocol is typed, it picks the right operation and arguments instead of guessing at a UI.
Install in one line
The server is published on both npm and Docker Hub, so you can run it however your setup prefers:
claude mcp add risal --scope user \
--env RISAL_API_KEY=risal_pat_… \
-- npx -y risal-mcp-server
Prefer containers? Swap the last line for docker run -i --rm -e RISAL_API_KEY shaxzod/risal-mcp-server:latest. Either way, authentication is a single personal access token scoped to exactly what you allow.
Capabilities
The MCP server covers the surface teams automate most: creating and updating tasks (individually or in bulk), parsing free text into structured tasks, managing projects and sprints, starting and stopping timers, logging time, reading AI-usage stats, and operating servers — listing hosts and metrics, reading Docker/PM2 logs, and restarting services.
Why an MCP server beats a webhook
A webhook fires one way and forgets. MCP is a live, two-way contract: the agent discovers what it can do, reads current state, acts, and reads the result back to decide the next step. That feedback loop is what turns “automation” into an assistant that can actually finish a multi-step task.
Safe by design
Everything runs through your token. Scope it narrowly, rotate it when you want, and point RISAL_API_BASE at a self-hosted instance if you run one. The agent never sees more than the token allows.
If you build with AI agents, the Risal MCP server is the shortest path from “ask” to “done” across your tasks and infrastructure.