More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Circuit Breaker Sentinel — Anti-Loop & Error Guard

    1

    Halt costly AI self-healing loops with strict 3-strike failure limit and mandatory root-cause rubber-ducking.

    $26

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Circuit Breaker Sentinel — Anti-Loop & Error Guard

    Circuit Breaker Sentinel — Anti-Loop & Error Guard

    Example session with this skill installed

    Run the integration test suite with Circuit Breaker Sentinel active. If the same assertion error repeats 3 times, trip the circuit and trigger root-cause deduction.

    • Read your context and instructions
    • Compiled the circuit breaker sentinel

    Circuit Breaker Sentinel Status: TRIPPED

    • Monitored Command: pytest tests/e2e/test_checkout.py
    • Detected Recurring Error: AssertionError: Stripe webhook signature mismatch
    • Consecutive Failures: 3 Strikes reached

    Strike History:

    1. Strike 1 (Attempt 1): Agent modified webhook parser logic -> Failed.
    2. Strike 2 (Attempt 2): Agent re-generated mock payload timestamps -> Failed.
    3. Strike 3 (Attempt 3): Agent bypassed signature verification in test -> Rejected by policy.
    4. 🛑 TRIP: Circuit opened. Execution halted to protect token budget ($4.20 saved).

    Mandatory Rubber-Duck Root Cause Deduction:

    The webhook secret in .env.test does not match the signing secret in conftest.py. No code modification permitted until the test environment secret is synced.

    Next Steps:

    Run: export STRIPE_WEBHOOK_SECRET=whsec_test123 and reset circuit breaker.

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

    About this skill

    Circuit Breaker Sentinel — Anti-Loop & Error Guard

    Tired of autonomous AI agents burning $10 to $50 in token loops trying—and failing—the same broken fix over and over again?

    When autonomous AI coding agents encounter runtime errors, test failures, or build crashes, their default behavior is a dangerous spiral: make a blind fix, rerun, fail, make another blind variation, repeat. This "infinite self-healing loop" burns your API budget and corrupts your codebase.

    ⚡ The Solution: The Deterministic 3-Strike Circuit Breaker

    The Circuit Breaker Sentinel enforces a strict, deterministic 3-strike policy with mandatory Rubber Duck root-cause deduction before allowing any further file modifications.

    🛡️ How the 3-Strike Protocol Works:

    1. Strike 1 (Surgical Diagnosis): The agent must state a formal hypothesis before touching code.
    2. Strike 2 (Premise Invalidation): If Strike 1 fails, the previous premise is invalidated. The agent must inspect surrounding context and dependencies.
    3. Strike 3 (Final Refactor): One final attempt with proof of root cause.
    4. 🛑 TRIP — Hard Cutoff: If unresolved after 3 attempts, the circuit breaker TRIPS. All tool execution is immediately halted to protect your token budget and codebase integrity. A 4th blind attempt is strictly prohibited without human signoff.

    📦 What's Included:

    • scripts/circuit_breaker_sentinel.py (CLI loop detector & strike evaluator)
    • references/circuit_breaker_protocol.md (Rubber-ducking templates and deduction rules)
    • agents/openai.yaml (Agent configuration & execution rules)
    • Ready-to-use hooks for Claude Code, Cursor, Windsurf, and Antigravity

    🔒 100% Local-First & Zero Telemetry

    Runs entirely on your local machine using your existing CLI environment. Zero data telemetry, zero cloud egress.

    Requires Python 3 environment. Relies on the agent's ability to track session history or logs to count strikes accurately.

    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 27 days ago

    Permissions

    Terminal / Shell
    Read Files
    Listed1 month ago
    Updated27 days ago

    What's inside

    Frequently Asked Questions

    The Sentinel acts as a 'kill switch' by monitoring task failures. It stops autonomous agents from spiraling into infinite, costly loops by enforcing a hard three-strike limit and requiring structured reasoning before any file edits are allowed.

    This skill is designed for autonomous coding environments like Claude Code and Cursor. It works within any setup where the agent has permission to execute Python scripts and read/write local session logs.

    Setup is straightforward: you'll receive a Python monitoring script and a set of integration instructions to include in your agent's system prompt or configuration file. No complex API keys or external platform subscriptions are required.

    You will receive the `circuit_breaker_sentinel.py` core script, the structured Markdown reasoning protocol, and a comprehensive integration guide to help you deploy the logic across your specific agent workflows.

    Yes, the sentinel requires a local Python runtime to track strike counts and a persistent session (or local log file) so the agent can reference its previous failed attempts across different turns.

    While standard prompts are easily ignored during high-velocity loops, this skill forces a logic gate at the execution layer. It requires the agent to pass a validation step before it can regain write-access to your codebase, effectively overriding the agent's tendency to repeat mistakes.

    New in Agents & Orchestration