Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    prooffix

    1

    Stop guessing on bug fixes. Forces your AI agent to prove the bug fails first, ship a minimal patch, audit diff hygiene, and deliver command-backed proof.

    $19

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    prooffix

    prooffix

    Example session with this skill installed

    Fix the bug where the calculate_total function in orders.py returns None instead of 0 for empty carts. Use pytest for the reproduction.

    • Read your context and instructions
    • Compiled the prooffix

    I have reproduced the defect where an empty cart caused a TypeError due to an unexpected None return. I implemented a surgical fix and verified it with a new regression test and the existing suite.

    Verification Report

    MetricDetails
    StatusFixed and verified
    Defectcalculate_total returns None for empty input
    Reproductiontests/reproduce_issue.py (Failed with TypeError)
    FixInitialized total to 0 at function start
    Hygiene0 debug statements found; 0 tests skipped

    Regression Test Output (Before Fix)

    $ pytest tests/reproduce_issue.py
FAILED tests/reproduce_issue.py::test_empty_cart_returns_zero - TypeError: unsupported operand type(s) for +: 'NoneType' and 'int'

    Verification Output (After Fix)

    $ pytest tests/reproduce_issue.py tests/test_orders.py
tests/reproduce_issue.py .                                               [ 50%]
tests/test_orders.py ....                                                 [100%]
PASSED [5/5 tests]

    Implementation Diff

    def calculate_total(items):
-    if not items:
-        return None
+    total = 0
     for item in items:
         total += item.price
     return total

    Next steps

    1. Review the tests/reproduce_issue.py file to decide if it should be merged into the main suite.
    2. Run the full integration suite to ensure no side effects in the checkout flow.
    3. Delete the temporary reproduction script before pushing.

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

    About this skill

    What it does

    ProofFix turns your AI coding agent into a disciplined, evidence-first debugging engineer. Instead of rushing to edit code and claiming "fixed," the agent must follow a strict 6-step protocol backed by three local verification scripts:

    1. Understand & Isolate: Reads repository rules, manifests, and failing paths before modifying a single line.
    2. Baseline First: Detects the repository's native test runner (scripts/detect_test_runner.py) and runs the narrowest existing check to separate pre-existing failures from new regressions.
    3. Prove Failure Before Fixing (Red): Writes a deterministic regression test targeting observable behavior and executes it against the unfixed code to prove it fails for the exact reported defect.
    4. Minimal Root-Cause Fix (Green): Implements the smallest contract-preserving change at the authoritative boundary — blocking broad exception-swallowing fallbacks, falsy-value bugs, or unrelated refactoring.
    5. Before-and-After Verification: Re-runs the regression test, module suite, and relevant lint/typecheck/build commands, clearly distinguishing verified checks from unrun ones.
    6. Diff Hygiene & Report Validation: Runs scripts/verify_diff_hygiene.py to catch newly added .skip/.only/@pytest.mark.skip, leftover debugger/breakpoint()/console.log, or lockfile churn, then validates the final report with scripts/check_report.py.

    Why use this skill

    Standard AI prompts frequently produce plausible patches with weak assertions, mocked-out bugs, or unexecuted claims. ProofFix enforces hard non-negotiable rules: no claimed verification without observed commands, no test weakening, no fabricated output, and explicit disclosure of unrun checks and remaining risks.

    What's inside (prooffix.zip)

    • SKILL.md — Trigger-tuned workflow instructions compatible across Claude Code, Cursor, Codex CLI, Gemini CLI, GitHub Copilot, Windsurf, OpenClaw, and 20+ SKILL.md agents.
    • agents/openai.yaml — Codex CLI display metadata.
    • scripts/detect_test_runner.py — Zero-dependency detector for JS/TS (npm/pnpm/yarn/bun, Vitest, Jest, Mocha, Playwright, node --test), Python (pytest/unittest), Go, Rust, and Java/JVM (Maven/Gradle).
    • scripts/verify_diff_hygiene.py — Read-only git diff & untracked file scanner blocking skipped tests, debug leftovers, and unexpected lockfile changes.
    • scripts/check_report.py — Schema validator ensuring every final verification report includes valid Status, Root Cause, Changes, Verification Evidence, Checks Not Run, and Remaining Risk.
    • references/debugging-playbooks.md — Targeted reproduction tactics for JS/TS, Python, Go, Java/JVM, Rust, intermittent/concurrency issues, and performance benchmarks.
    • references/anti-patterns-and-guardrails.md — Concrete AI shortcut traps vs. ProofFix standards.
    • references/verification-report-template.md — Structured report template.
    • examples/ — Real executed Python (pytest) and JavaScript (node --test) walkthroughs with actual terminal outputs.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Testing & Debugging.

    Requires an existing local test runner. Cannot verify fixes if the environment or dependencies prevent execution. Skips UI visual-only guesses without DOM/E2E tools.

    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 buying a skill.

    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 today

    Software
    vitestNode.jsjestPlaywrightPythonmocha

    Permissions

    Terminal / Shell
    Browser
    Read Files
    Write Files
    Listedtoday

    What's inside

    Frequently Asked Questions

    prooffix eliminates "plausible but broken" code by enforcing a strict Test-Driven Development (TDD) cycle. It mandates that a failing test case be successfully executed before any production code is modified, preventing the common AI habit of guessing at a solution.

    The skill includes native support for JavaScript/TypeScript (node --test), Python (pytest, unittest), Go, Java/JVM, and Rust. It uses bundled discovery scripts to automatically detect your environment's specific test runner and project structure.

    You receive the core logic for the "no reproduction, no fix" workflow, specialized scripts for diff hygiene and debug statement detection, and structured reporting templates for verification logs. It is compatible with major AI coding agents and IDE extensions that support custom skill integration.

    prooffix is designed to integrate into your existing local development environment or CI pipeline. It requires access to your local workspace and test runners, but does not need access to external production systems or unauthorized APIs to function.

    Beyond the fix itself, prooffix performs "diff hygiene" to scrub your code of debug statements like console.log or pdb. It also generates a canonical verification report confirming that the regression test passed without weakening existing assertions.

    New in Testing & Debugging