openkoi-ai

Guides OpenKoi when working on its own codebase. Enforces architectural invariants, zero-dependency crypto, test-first iteration, and safe self-modification patterns.

Evaluates API endpoint design for RESTfulness, consistency, error handling, and documentation.

Evaluates code changes for correctness, style, and safety.

General-purpose evaluator for any task type. Used as fallback when no domain-specific evaluator matches.

Evaluates written content for clarity, accuracy, tone, and structure.

Evaluates SQL queries and database migrations for correctness, safety, performance, and reversibility.

Evaluates test code for coverage, assertion quality, isolation, and readability.