cursor-cloud-agents
by parcostabot
Coding Agents & IDEsBrief
Deploy Cursor AI agents to GitHub repos.
Description
What This Skill Does
Wraps the Cursor Cloud Agents HTTP API so OpenClaw can dispatch coding tasks to Cursor's cloud, monitor agent progress, and retrieve results. Agents run against GitHub repositories and can write code, generate tests, produce documentation, and open PRs. Supports fire-and-forget, supervised, iterative, and background execution patterns.
Offloads long-running coding tasks to Cursor's cloud so local dev work is not blocked and no local environment setup is needed for the target repo.
When to Use It
- Generating unit tests for a specific module without blocking local work
- Opening a PR with a new feature implemented by a cloud agent
- Running a large codebase refactor overnight in background mode
- Sending follow-up prompts to refine code an agent already wrote
- Checking quota usage before launching a batch of coding tasks
Example Workflow
User asks: Add unit tests for the auth module in owner/myrepo
- 1
Verifies repository access with cursor-api.sh verify owner/myrepo
- 2
Launches agent with --repo owner/myrepo and a specific test-writing prompt
- 3
Polls cursor-api.sh status until agent reports FINISHED
- 4
Retrieves conversation history to confirm PR URL
Agent opens a pull request containing unit tests for the auth module
Requirements
- CURSOR_API_KEY from Cursor IDE Settings > General
- Cursor subscription (Pro or Ultra for concurrent agent access)
- Cursor GitHub App installed on target repositories
Installation
clawhub install parcostabot/cursor-cloud-agentsRun the CLI command, or paste the prompt into Claude Code, Cursor, Codex or whichever coding agent you use.