MCP Server for Project Management: Connect Claude and Other Agents
Setup takes two steps: create a scoped API key, then add one server entry to your AI client. This is the whole procedure, with the security choices that matter.
To connect Claude or another agent to CoMng.AI over MCP, create a scoped API key as an Admin, then add the server URL https://comng.ai/app/api/external/mcp with an X-API-Key header to your client (Claude Code, Claude Desktop, Codex CLI, Gemini CLI or Antigravity), and test with a read-only request first.
- Step 1: an Admin creates a scoped API key in API Management.
- Step 2: add the MCP server URL and the X-API-Key header to your AI client.
- Least privilege: tick only the actions the agent needs; add an IP whitelist and an expiry date.
- Test first: the API sandbox lets you send requests and see responses, including cURL commands, before wiring an agent.
Step 1: Create a Scoped API Key
Only Admins can create API keys. In API Management, choose Create API Key and set:
- Name and description so you can recognise the key later, for example "Claude Code agent".
- Projects: choose specific projects rather than all, unless the agent genuinely needs everything.
- Access level and actions: start with the minimum. A planning agent usually needs list and update for tasks and notes, and create for activities, but not delete.
- Rate limit: lower it for scheduled jobs.
- Expiry and IP whitelist: for a fixed-IP host such as a CI server, an IP whitelist is the single most effective protection against a leaked key.
The full key is shown once. After you close the dialog only a masked value is stored. If you lose it, revoke it and create another.
Step 2: Connect Your AI Client
| Client | How to add the server |
|---|---|
| Claude Code | Run claude mcp add --transport http comng-ai https://comng.ai/app/api/external/mcp --header "X-API-Key: <key>" |
| Claude.ai / Claude Desktop | Settings → Connectors → Add custom connector; enter the URL and the X-API-Key header |
| OpenAI Codex CLI | Add an [mcp_servers.comng-ai] block with the URL and a bearer token to ~/.codex/config.toml |
| Gemini CLI | Add an mcpServers.comng-ai entry with httpUrl and an X-API-Key header to ~/.gemini/settings.json |
| Google Antigravity | Add it under MCP Servers settings (or mcp_config.json) the same way |
Step 3: Verify Before You Trust It
- In the API sandbox, select your key, pick an endpoint such as list tasks and send a request. You see the status code, response time, JSON body and ready-to-copy cURL and wget commands.
- In your AI client, ask something read-only: "List the overdue tasks in the project."
- Then try a narrow write: "Add an activity noting that setup is complete."
- Check the Audit Logs and the key's usage counters in API Management.
If something returns 401, the key is wrong or revoked; 403 means the request came from an IP outside the whitelist; 429 means the rate limit was exceeded.
What to Ask Your Agent First
- "Which tasks are blocked by overdue predecessors?" It can use the dependency tools to answer from live data.
- "Which milestones depend on tasks scheduled after their due date?" It reads milestone dependencies and dates.
- "Generate the project Pulse and summarise the top three alerts."
- "Draft a change request that moves Milestone X to match the plan." Applying still needs a person.
Background on the protocol is in The MCP Bridge, and a full agent workflow in Claude Code Project Management.
- Pasting exports into a chat window
- One shared all-powerful key
- No idea what the agent did
- The agent reads the live project
- A narrow key per agent, revocable any time
- Usage counters and audit logs
About CoMng.AI
CoMng.AI is an AI-native project management platform that generates a project's plan, tasks, risks and budget from a description or document, monitors project health in real time, and gives both human teams and AI coding agents live access to the same project data, with a person approving every AI-generated change.
Give your AI agents a live plan and a safe way to work in it
Create a scoped API key, connect your agent over MCP and let it pick up work from the same project your team uses.
Frequently Asked Questions
How do I connect Claude to my project management data?
Create a scoped API key in CoMng.AI API Management, then add the MCP server URL https://comng.ai/app/api/external/mcp with an X-API-Key header as a custom connector in Claude.ai or Claude Desktop, or with claude mcp add in Claude Code.
Do I need to be an Admin?
Creating and viewing API keys is restricted to Admins. Project Managers and Members can use an integration once an Admin has issued a key.
What if my key leaks?
Disable or revoke it immediately in API Management. Prevent misuse in advance with an IP whitelist, an expiry date and minimal action permissions.
Can I use it with OpenAI or Gemini tools?
Yes. OpenAI Codex CLI, Gemini CLI and Google Antigravity are supported through their MCP configuration, as is any MCP-compatible client.
Can the agent delete my data?
Only if the key allows it. Leave delete actions unticked unless the agent genuinely needs them.
Related reading
See the full guide: AI Agents, ARI & MCP Guide