Forge

What it does

Forge is an autonomous behavioral validation swarm skill for Claude Code that uses Gherkin specifications as the single source of truth. It employs eight specialized agents working in parallel to continuously verify, test, fix, and commit code until all Gherkin scenarios pass and seven behavioral validation gates (Functional, Behavioral, Coverage, Security, Accessibility, Resilience, Contract) are cleared. Forge leverages confidence-tiered fix patterns and defect prediction based on historical data for automated bug resolution.

When to use it

• Automated Regression Testing: Ensure new code changes don't break existing functionality by automatically retesting all scenarios after each commit.
• Behavioral Validation in Microservices Architectures: Verify interactions between multiple services without manual integration testing.
• Security and Accessibility Audits: Continuously validate security and accessibility requirements as part of the development process.
• Monorepo Testing: Streamline testing across a large codebase with shared types and cross-cutting validation.

Key capabilities

• 8 specialized agents working in parallel
• Gherkin behavioral specifications
• 7 behavioral validation gates (Functional, Behavioral, Coverage, Security, Accessibility, Resilience, Contract)
• Confidence-tiered fix patterns (Platinum/Gold/Silver/Bronze)
• Defect prediction
• Chaos/resilience testing
• Visual regression testing
• External-only mocking

Example prompts

• "Forge, run all Gherkin scenarios and report any failures."
• "Forge, focus on the 'Security' validation gate for the payment processing module."
• “Forge, perform mutation testing on the user authentication component.”

Tips & gotchas

Forge is designed for Claude Code. Ensure you have a compatible environment set up before attempting to use it. The system relies heavily on Gherkin specifications; ensure these are well-defined and accurate for optimal results.