reckit

What This Skill Does

Reckit is an advanced, autonomous code verification engine designed for OpenClaw. It shifts AI development from speculative coding to evidence-based engineering by running a multi-stage verification pipeline. Instead of trusting a single model's output, Reckit spawns parallel workers to act as builders, testers, and breakers. It enforces a rigorous 14-step framework that includes slop-scanning for low-quality code, strict type-checking, mutation testing to verify test efficacy, and cross-verification against independent oracles. The skill produces a formal '.wreckit/' proof bundle, which contains the gate results and a definitive Ship, Caution, or Blocked verdict, ensuring that code is production-ready before deployment.

Installation

To integrate Reckit into your OpenClaw environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/christiancattaneo/wreckit-ralph Once installed, ensure your project environment has the necessary permissions for file I/O, as the agent requires access to read your source files and write the proof bundle directory.

Use Cases

Reckit is designed for high-stakes software engineering. Use it when:

1. Building new greenfield projects where reliability is non-negotiable.
2. Migrating legacy codebases, such as porting existing logic into TypeScript or Swift.
3. Resolving critical bugs while mathematically ensuring that no regressions are introduced in existing functionality.
4. Performing deep security and quality audits on external or inherited codebases.
5. Implementing robust CI/CD gate checks where automated proof of test quality is required.

Example Prompts

1. "Build me a secure user authentication module in React using the reckit verification framework to ensure zero vulnerability."
2. "Audit this legacy backend project, perform a full mutation test, and generate a proof bundle to identify any silent code rot."
3. "Fix the circular dependency bug in the auth service; verify the fix with a regression test and provide a proof bundle for the changes."

Tips & Limitations

• Gate Selection: Reckit auto-detects your mode (Build, Rebuild, Fix, Audit). You can force a specific gate by mentioning the mode in your prompt.
• Computation Costs: Because Reckit uses parallel agents and mutation testing, it is more compute-intensive than standard code generation.
• No Rewriting: Do not modify the core Reckit scripts located in the source repo; these are deterministic and essential for the integrity of the proof bundle.
• LLM-as-Judge: For the most critical paths, enable the 'LLM-as-Judge' gate to provide a semantic review of your logic in addition to the automated testing.