About CodeChecks

Verified, policy-enforced AI code generation you can trust.

CodeChecks takes developer requirements and produces trustworthy, auditable code patches. Every run follows a structured workflow: plan, implement, verify. The result is a unified diff you can inspect, apply, and reproduce — backed by an evidence bundle that records exactly what happened.

What makes CodeChecks different

Repo-aware generation

CodeChecks introspects your project structure, dependencies, scripts, and test framework before generating code — so the output matches your conventions.

Policy enforcement

Dependency files, lockfiles, and sensitive files (env vars, auth, payments) are protected by default. Changes require explicit opt-in via policy flags.

Evidence bundles

Every run produces a JSON evidence record with repo SHA, patch hash, verification commands, environment info, and step-by-step reproduce instructions.

Provider resilience

Automatic retries with exponential backoff, rate-limit handling, and transparent fallback to a secondary model if the primary provider fails.

How it works

  1. 1
    Submit requirements — describe what you want to build, your tech stack, and constraints.
  2. 2
    Review the plan — the Planner produces a structured task list with file plans and verification commands.
  3. 3
    Implement tasks — the Implementer generates a unified diff for each task, policy-checked before delivery.
  4. 4
    Verify — run verification commands, submit results, and get an automated quality report.