Integration · Developer tools

Connect Claude & ChatGPT to Cursor Cloud Agents

Ask your AI to hand a coding task to a Cursor Cloud Agent, check how it is going, or send it a follow-up. It can launch agents on your GitHub repositories, pick a model, open a pull request when the work is done, and read runs, artifacts and usage. You can start and steer background coding work from the chat you are already in.

Connect Cursor free

Free plan, no credit card. Takes about a minute.

Try asking

“Have a Cursor agent fix the failing date parser test in our api repo and open a PR.”
The AI launches a Cloud Agent on the repository with your task and auto_create_pr on. It runs on Cursor's machines and uses your Cursor usage.
“Which repositories can the agents work on?”
It lists the GitHub repositories your Cursor account can give Cloud Agents access to.
“How is the agent I started this morning doing?”
It lists your agents, opens the one you mean and reads the status of its latest run.

What your AI can do

Launch agents

Start Cloud Agents on a GitHub repository or pull request, choose a model and have them open a pull request when done.

Follow and steer

List agents and runs, read status and results, send follow-ups or ask for a plan, and cancel a run.

Output and usage

List and download the files an agent produced, and read token and cost usage per agent or run.

“Ask it to also add a test for leap years.”
It sends a follow-up instruction to the agent, which starts a new run.
“Before changing anything, ask the agent for a plan to split the billing module.”
It sends the follow-up in plan mode, so the agent returns a plan instead of code changes.
“What did that agent cost?”
It reads the agent's usage in tokens and cost, for all runs or one run.

15 tools for Cursor

These tools are switched on when you connect. You can switch any of them off, or require your approval before it runs.

archive_agentVisible to othersArchive an agent by `agent_id` to hide it from the active list. It can be restored later.
cancel_agent_runDeletesStop a run that is still working, by `agent_id` and `run_id`. Work it has not finished is lost.
download_agent_artifactDownload one artifact file from an agent by `agent_id` and its relative `path` under artifacts/.
get_agentGet one Cloud Agent by `agent_id`, including its status, repository, branch and pull request.
get_agent_runGet one run of an agent by `agent_id` and `run_id`, including its status and result.
get_agent_usageGet usage (tokens and cost) for an agent by `agent_id`, optionally for one `run_id`.
get_api_key_infoGet details about the connected Cursor API key and the user it belongs to. Use to confirm which account you are working in.
launch_agentChanges dataLaunch a Cloud Agent that works on code by itself: pass `prompt.text` with the task and usually `repos` with a GitHub repository `url` (optionally `startingRef` or a `prUrl`). Optionally pick `model.id` and set `auto_create_pr` to open a pull request on GitHub when done. It runs on Cursor's machines and uses Cursor usage.
list_agent_artifactsList the files an agent produced (artifacts) by `agent_id`, with their paths.
list_agent_runsList the runs (each prompt and follow-up) of an agent by `agent_id`.
list_agentsList Cloud Agents with their IDs, names and status. Filter by `pr_url`, or set `include_archived` to include archived agents.
list_modelsList the AI models available for Cloud Agents, with the IDs to pass as `model.id` when launching an agent.
list_repositoriesList the GitHub repositories the Cursor account can give Cloud Agents access to, with their URLs.
send_agent_followupChanges dataGive an existing agent a follow-up instruction by `agent_id` with `prompt.text`, starting a new run. Set `mode` to plan to get a plan instead of code changes.
unarchive_agentChanges dataRestore an archived agent by `agent_id` so it shows in the active list again.

Set up in three steps

1
Pick the app
Create a free PipMCP account and choose Cursor from the app list.
2
Paste your key
Paste a Cursor API key. Open the Cursor Dashboard > API Keys (cursor.com/dashboard/api) and create a user API key, or use a service account key from team settings. Copy it right away, as it is shown only once.
3
Add the link to your AI
You get a personal MCP link. Add it to Claude, ChatGPT or Cursor:
1. Click your name, then Settings › Connectors › Add custom connector.
2. Paste your link as the Remote MCP server URL.
3. Switch it on from the + menu in a chat.

Questions

What can the AI do in Cursor?

The AI can list and read Cloud Agents, their runs, artifacts and usage, your available models and the GitHub repositories the account can reach. It can launch agents, send follow-ups, archive and restore agents, and cancel a run that is still working.

Does the AI see my Cursor credentials?

No. Your Cursor credentials are encrypted at rest and never shown to the AI. After you save them, they are not shown again, not even to you. The AI only sees the results of the tools it calls.

Can I control what the AI is allowed to do?

Yes. You choose which tools are switched on, so you can start by only following agents. Launching an agent uses Cursor usage and can open a pull request, and cancelling a run loses unfinished work. Both can require your approval before they run. Every tool call is logged.

Does it work with ChatGPT?

Yes. In ChatGPT, open Settings › Apps & Connectors › Advanced, turn on Developer mode and add your PipMCP link. Developer mode needs a paid ChatGPT plan: Plus, Pro, Business or Enterprise. The same link also works in Claude (Settings › Connectors › Add custom connector), Cursor and other MCP clients.

What does it cost?

PipMCP has a free plan with no credit card required. Paid plans bill per completed task. You also need your own Cursor account.

Does launching an agent use my Cursor usage?

Yes. Cloud Agents run on Cursor's machines under your Cursor account, so their work counts toward your Cursor usage. The AI can read the tokens and cost of each agent or run, and you can require approval before any agent is launched.