Cursor Agent Superpowers

    1

    An AI coding orchestration workflow where Codex plans, delegates, reviews, and verifies while Cursor Agent handles implementation.

    Free

    1 useSecurity scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Cursor Agent Superpowers

    Example session with this skill installed

    I have an approved plan with three tasks. Task 1 is a new API endpoint, Task 2 and 3 are independent UI components. Start the execution loop and tell me how you will handle the isolation.

    • Read your context and instructions
    • Compiled the cursor agent superpowers

    I have initialized the execution engine and checked the integration ledger at .superpowers/cursor-execution/progress.md. Since Task 1 (API) is a prerequisite for the UI components, I will execute Task 1 sequentially. Once approved, I will evaluate Tasks 2 and 3 for a parallel wave based on their write sets.

    Execution Ledger Status

    Task IDStatusWorktree PathBranchHEAD
    task-001pending/worktrees/task-001feat/api-endpointa1b2c3d
    task-002queuedTBDfeat/ui-headerpending
    task-003queuedTBDfeat/ui-footerpending

    Execution Plan:

    1. Dispatch Task 1: I am creating a fresh Cursor session via cursor-agent-bridge for the API implementation.

    Review Gate: After the worker signals implementation, I will trigger reviewing-cursor-changes to verify the endpoint against the spec.
    3.

    Wave Evaluation: If Task 2 and 3 prove to have disjoint write sets (e.g., different React components) and share no integration surface, I will dispatch them into sibling isolated worktrees.

    Next steps

    • Monitor stdout.txt for the Task 1 bridge run.
    • Review the brief.md generated in the task artifact root.
    • Validate the API schema before authorizing the parallel UI wave.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Cursor Agent Superpowers

    Plan with Codex. Build with Cursor. Verify with evidence.

    Cursor Agent Superpowers is a free, open-source AI coding orchestration workflow designed to coordinate Codex and Cursor Agent CLI through structured planning, task delegation, implementation, code review, and verification.

    Instead of letting a single AI agent handle everything, it separates responsibilities between a controller and an implementation agent.

    Key Features

    • Agent Orchestration: Codex acts as the controller, while Cursor Agent handles implementation tasks.
    • Structured Planning: Break development work into clearly scoped, reviewable tasks.
    • Controlled Execution: Execute tasks sequentially by default, with dependency-proven parallel execution when safe.
    • Automated Task Delegation: Dispatch bounded tasks through a Python-based Cursor Agent bridge.
    • Evidence-Based Review: Verify actual code changes, execution records, test results, and Git state before approving work.
    • Git Worktree Isolation: Keep implementation work isolated and maintain clear task boundaries.
    • Optional Superpowers Integration: Work with upstream Superpowers skills when available, or use controller-native workflows.

    Included Skills

    1. using-cursor-superpowers — Workflow routing and coordination.
    2. cursor-agent-bridge — Cursor Agent CLI task execution and evidence capture.
    3. executing-plans-with-cursor — Controlled execution of approved implementation plans.
    4. reviewing-cursor-changes — Code review and verification gates.

    All four components are designed to work together.

    Who Is It For?

    • Developers using Codex and Cursor Agent together.
    • AI-assisted software engineering teams.
    • Developers building multi-agent coding workflows.
    • Engineers who want stronger review and verification controls.

    Requirements

    • Cursor Agent CLI with valid access
    • Python 3
    • Git

    The bridge currently uses a configured Cursor model and requires a working local Cursor Agent environment.

    Open Source

    100% free and open-source under the MIT License.

    GitHub: https://github.com/hunterT20/cursor-agent-superpowers

    For the full installation procedure, dependencies, and latest updates, please refer to the GitHub repository.

    Requires Python for bridge scripts and git worktree support. Cannot auto-resolve merge conflicts or auto-rollback history. Maximum three concurrent tasks in parallel waves.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    1 use

    Times used to date

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 27 days ago

    • Free with an Agensi account

    Requires

    Node.js
    Python

    Permissions

    Read Files
    Write Files
    Listed27 days ago

    What's inside

    Frequently Asked Questions

    It prevents race conditions and corrupted git history by using isolated worktrees and a structured integration ledger to ensure tasks are executed and verified in the correct order.

    This skill is purpose-built for the Cursor IDE and its native Agent mode, requiring a local Python environment and Git to manage the bridge scripts and worktrees.

    You will receive the structured implementation bridge scripts, the ledger configuration templates, and a comprehensive guide on integrating the workflow into your Cursor project.

    No, the skill includes a Python-based bridge and automated ledger tracking that maintains the state of your implementation plan even if you restart your session or clear your chat history.

    Yes, the skill supports parallel execution via isolated git worktrees, provided the tasks pass an initial validation check to ensure they don't share conflicting APIs or schemas.

    New in Agents & Orchestration