Run Cursor agents as a team in Superconductor
Connect your Cursor API key to Superconductor and your whole team can run Cursor agents on cloud computers everyone can watch. With Superconductor, your team can run Composer 2.5 next to Claude Code on the same ticket and review the results together.
Superconductor lets you and your team chat with coding agents such as Claude Code, Codex, Pi, and more in shared, secure cloud sessions.
With our Cursor integration, you can
- Run Composer 2.5 next to Claude Code and other agents on one ticket
- Benchmark Composer models on your merged PRs
- Use previews, automations, and most workspace connectors
Technical details
- Runs on
- A Cursor API key
- Models
- Composer 2.5, OpenAI, Anthropic, Google, xAI models, or Auto
- Runs as
- A Superconductor cloud computer, not a Cursor Cloud Agent
Example
Questions
- Is this the same as Cursor Cloud Agents?
- No. Your Cursor API key gives Superconductor access to Cursor's models and agent, which run on a Superconductor cloud computer. Cursor's own automations and reporting don't apply.
- Does every MCP server work with Cursor?
- Almost. With Composer 2.5, Composer 2.5 Fast, the Grok profiles, and Auto, Superconductor leaves out the PostHog MCP server for now because of a compatibility issue. Every other configured server still works.
- Do I need other API keys to benchmark Cursor?
- No. The Cursor key is enough, and Superconductor covers the judge models that score the results. If your project or workspace has an Anthropic, OpenAI, or Gemini key, that judge uses it instead.