cursor-for-claude

agent
Security Audit
Fail
Health Warn
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 7 GitHub stars
Code Fail
  • process.env — Environment variable access in plugins/cursor/scripts/cursor-companion.mjs
  • process.env — Environment variable access in plugins/cursor/scripts/lib/cursor.mjs
  • spawnSync — Synchronous process spawning in plugins/cursor/scripts/lib/process.mjs
  • process.env — Environment variable access in plugins/cursor/scripts/lib/process.mjs
Permissions Pass
  • Permissions — No dangerous permissions requested
Purpose
This tool acts as a bridge between Claude Code and your local Cursor environment. It allows developers to delegate coding tasks to Cursor in the background and track their completion status without leaving the Claude interface.

Security Assessment
The overall risk is Medium. The plugin operates by executing shell commands directly on your machine, utilizing synchronous process spawning to run the `cursor-agent` CLI in the background. It relies heavily on environment variables, specifically looking for a `CURSOR_API_KEY` and standard paths to function. While no dangerous global permissions or hardcoded secrets were detected, the fundamental nature of the tool involves taking a prompt and executing it via command line. You must inherently trust the tool because it has the capability to run arbitrary commands and scripts within your local workspace.

Quality Assessment
The project uses a standard MIT license and is considered actively maintained, with very recent updates pushed to the repository. However, community trust and visibility are extremely low. The project currently has only 7 GitHub stars, meaning it has not been widely vetted or battle-tested by a large audience. Developers should expect limited community support or documentation if issues arise.

Verdict
Use with caution. Ensure you review the codebase yourself before executing, as low-visibility tools with local command execution capabilities carry inherent risks.
SUMMARY

Execute plans and delegate tasks to Cursor in the background from Claude Code

README.md

cursor-plugin-cc

A Claude Code plugin that delegates coding tasks to your local cursor-agent and tracks background jobs per repository.

Requirements

  • Claude Code (with plugin support)
  • Node.js on PATH
  • cursor-agent installed and authenticated (CURSOR_API_KEY or cursor-agent interactive login)

Install

Add this folder as a local marketplace in Claude Code, enable the cursor plugin, then run:

/cursor:setup

/cursor:setup checks the CLI and auth, and can install cursor-agent for you if missing.

Commands

Command What it does
/cursor:agent <task> Delegate a task to Cursor (default: background, write-capable)
/cursor:sessions [id] List jobs for this repo, or show one
/cursor:done [id] Show the stored final output for a finished job
/cursor:stop [id] Cancel an active background job
/cursor:setup Check install + auth

Examples

/cursor:agent fix the failing parser tests
/cursor:agent --wait quick rename in this file
/cursor:agent --read-only investigate why checkout is slow
/cursor:agent --continue apply the next fix
/cursor:agent --mode plan map the auth refactor

Default flow (background):

/cursor:agent <task>
/cursor:sessions
/cursor:done <job-id>

Use --wait when you want the result inline.

How it works

The plugin spawns cursor-agent -p "<prompt>" --output-format stream-json --workspace <repo> and parses the JSON stream. Job state lives under CLAUDE_PLUGIN_DATA (or a temp cursor-companion directory as fallback).

License

MIT.

Reviews (0)

No results found