Blind Auditor

소개

Blind Auditor is an innovative tool designed to enhance the quality and compliance of code generated by AI agents. It implements a 'thought isolation' mechanism, compelling AI agents to perform self-reviews after code generation by switching to an impartial auditor role. The tool integrates seamlessly with existing IDEs and leverages host inference models, eliminating the need for additional API keys. It hard-embeds team-defined coding standards and security policies, automatically triggering code corrections upon audit failure and ensuring that its rules take absolute priority over conversational prompts.

주요 기능

• Strong compliance by hard-embedding team coding standards via `rules.json`.
• Zero-cost operation by reusing the host IDE's inference model.
• Bias reduction through forced agent role-switching to an 'auditor'.
• Automated correction loop for audit failures, with configurable retries.
• High enforcement priority for MCP rules over conversational prompts.
• 0 GitHub stars

사용 사례

• Automating code quality and security checks for AI-generated code.
• Enforcing organizational coding standards and policies in AI development workflows.
• Integrating mandatory self-review processes for AI agents creating software.