Cursor self hosted agent

Why deploy cursor self hosted agent on Render?

Cursor self-hosted cloud agents is a deployment model that runs Cursor's AI coding agents on your own infrastructure while Cursor handles orchestration and model access. It solves the problem of keeping code execution and tool access within your private network without requiring inbound ports, VPN tunnels, or Kubernetes management.

This template wires together a Cursor self-hosted worker and trigger API with pre-configured service communication, so you skip the manual setup of connecting background workers to webhook endpoints and the Cursor control plane. Instead of configuring worker lifecycle management, outbound networking, and multi-entrypoint routing yourself, you get a working Slack/Linear/HTTP trigger pipeline with one click. Render's background workers handle the long-lived agent sessions without Kubernetes overhead, and you can iterate on the trigger logic using preview environments before promoting changes.

Architecture

What you can build

After deploying, you'll have a self-hosted Cursor cloud agent running on Render that executes coding tasks in your own infrastructure while Cursor handles orchestration. You can trigger agent runs via Slack commands, Linear webhooks, or direct HTTP requests—each kick-off returns an agent URL immediately and posts updates when the work completes. The worker clones your specified repository, so agents can read code, make changes, and open pull requests against it.

Key features

• Outbound-only HTTPS worker: The Cursor worker connects outbound over HTTPS to Cursor's control plane, requiring no inbound ports, VPN tunnels, or Kubernetes.
• Multi-source trigger API: A Node.js service normalizes incoming requests from Slack slash commands, Linear webhooks, and direct HTTP into Cursor Cloud Agent launches.
• Render Blueprint deployment: Ships as a ready-to-apply Render Blueprint that provisions both the trigger web service and background worker from environment variables.
• Single-repo isolation model: Each deployment is pinned to one target repository, with the worker cloning it at startup and the trigger API rejecting mismatched requests.
• Webhook completion callbacks: Supports posting agent completion updates back to Slack threads or Linear issue comments when API keys and webhook secrets are configured.

Use cases

• Platform team deploys AI coding agents without managing Kubernetes infrastructure
• Developer triggers bug fixes from Slack slash commands automatically
• Engineering manager routes Linear issues to AI agents for triage
• Startup runs secure code generation within their own network boundaries

What's included

Prerequisites

• Cursor API Key: API key used to authenticate with the Cursor Cloud Agents API for launching and managing agents.
• Cursor Target Repository: The Git repository URL that the worker will clone and that agents will operate against.
• Cursor Webhook Secret: A shared secret you create to verify incoming webhook callbacks from Cursor.
• Cursor Git Token: A personal access token or deploy key for cloning, fetching, and pushing to private repositories.
• OPTIONAL Slack Signing Secret: Secret from your Slack app used to verify incoming slash command requests.
• OPTIONAL Linear Webhook Secret: Secret used to verify incoming webhook requests from Linear.
• OPTIONAL Linear API Key: API key for posting comments back to Linear issues when agent runs complete.

Next steps

1. Open the Cursor Cloud Agents dashboard and verify your worker — You should see the Render-hosted worker listed as connected and available for sessions
2. Test the trigger API health endpoint at https://.onrender.com/healthz — You should receive a JSON response with ok: true confirming the service is ready
3. Send a test task via curl to /v1/tasks with a simple prompt like 'List the files in the repo' — You should receive an agent URL in the response and see the task appear in your Cursor Cloud Agents dashboard within a few seconds

Resources

• Cursor official site