Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Autonomous YOLO Harness & Catastrophic Safeguard

    1

    Autonomous execution mode with 0.15s prompt auto-approvals and hardcoded catastrophic safety guardrails against filesystem wipes and destructive commands.

    $2.99

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Autonomous YOLO Harness & Catastrophic Safeguard

    Autonomous YOLO Harness & Catastrophic Safeguard

    Example session with this skill installed

    Refactor the entire authentication middleware, run the automated test suite, and clean up temporary build artifacts.

    • Read your context and instructions
    • Compiled the autonomous yolo harness

    Execution Summary

    Autonomous YOLO Mode activated for current task session (`task` scope).
    
    | Phase | Actions Taken | Time Elapsed |
    | :--- | :--- | :--- |
    | **File Refactor** | Updated `auth.ts`, `session.ts`, and router middleware | 1.8s (Auto-Approved) |
    | **Test Suite** | Executed `npm test` across 14 test suites | 4.2s (Auto-Approved) |
    | **Cleanup** | Removed temporary `.cache` files | 0.3s (Auto-Approved) |
    
    
    ```text
    [YOLO Engine] ⚡ Auto-Approved command: `npm test` (0.14s)
    [YOLO Engine] ⚡ Auto-Approved edit: `src/middleware/auth.ts` (0.11s)
    [YOLO Engine] 🛡️ Catastrophic Gate: 0 destructive patterns detected.
    

    Next steps
    • All test suites passing (14/14).
    • Task completed: YOLO Mode has automatically disengaged and returned the terminal to guarded Human-in-the-Loop mode.

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

    About this skill

    Eliminate Interactive Bottlenecks Without Sacrificing System Safety

    When running autonomous coding agents on large refactors, repetitive confirmation modals (*"Run this command?"*, *"Allow editing this file?"*) break agent momentum and waste developer time. Completely unshielded flags, on the other hand, risk accidental filesystem damage.
    
    **Autonomous YOLO Harness** provides high-speed execution combined with hardcoded safety guardrails:
    
    * ⚡ **High-Speed Execution:** Intercepts confirmation prompts and approves them in 0.15s to keep workflows moving autonomously.
    * 🛡️ **Catastrophic Security Gate:** Hardcoded regex intercepts and pauses execution for manual review if a command involves root deletions (`rm -rf /`), disk formatting (`mkfs`), fork bombs, or destructive git force-pushes.
    * 🎯 **Session vs. Task Scopes:** Run indefinitely in `session` mode, or enable `task` mode to automatically revert to guarded mode once the current objective is completed.
    * 🔒 **Isolated Session State:** Enabling autonomous mode in one workspace or terminal session never leaves other sessions unguarded.
    * 🔄 **Zero-Failure Auto-Reset:** Sessions always start in guarded mode by default.
    

    Requires tmux and Python 3.10+ on a Linux, macOS, or WSL2 system. Operates via tmux buffer inspection. Built-in security rules specifically block known catastrophic patterns (e.g. root deletion, disk formatting, force pushes); custom forbidden commands can be easily added to the safety pattern list.

    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

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 17 days ago

    Requires

    Python
    Git
    tmux

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Environment Variables

    File Scopes

    ~/.gemini/**
    **

    The harness runs locally inside tmux to scan active terminal confirmation modals and auto-approve safe operations via keystroke injection. It accesses ~/.gemini/ for session state persistence and engine logging. No outbound network requests, external APIs, or data telemetry are used.

    Listed17 days ago

    What's inside

    Frequently Asked Questions

    This skill bypasses the repetitive manual "allow" prompts for every terminal command, allowing your agent to execute complex, multi-step tasks like code refactors autonomously while you are away from your keyboard.

    It is specifically designed for the Antigravity agent framework and requires a Unix-like environment with the Tmux-based background engine installed to function correctly.

    Even in autonomous mode, the skill features hard-coded safety rails that intercept and block catastrophic operations, such as forced git pushes or recursive root-level deletions.

    The 'task-specific' mode enables autonomy only for the current request and automatically reverts to guarded mode once the task is finished, ensuring you don't leave your system open indefinitely.

    You will receive the complete standalone autonomous harness package: the SKILL.md specification, background tmux supervisor daemon (yolo_tmux_engine.py), the session state controller (toggle_yolo.py), process watchdog, and setup documentation.

    Yes, it automatically guards git workflows by intercepting and blocking dangerous git force-pushes (--force, -f) and remote branch deletions while allowing safe commits and fast-forward pushes.

    New in Agents & Orchestration