intent-aware code review

Intent-Aware Code Review Guardrails

Intent-aware code review gives every pull request a memory of the rules your team already agreed to follow.

Intent-aware code review checks code changes against approved intent from CLAUDE.md, READMEs, ADRs, and architecture notes. AxiomGuard turns those approved rules into deterministic TypeScript checks with evidence a reviewer can audit.

Why it matters

Proof beats plausible review comments.

Candidate rules can come from documentation, but a human approves enforcement.

The YAML rule file remains the enforceable source of truth.

Findings include the violated invariant, changed file, evidence, status, and severity.

Use cases

Where teams apply this first.

FAQ

Questions about intent-aware code review.

What does intent-aware code review check?

It checks whether a change violates approved intent, including layer boundaries, forbidden side effects, and value invariants.

Does intent-aware code review rely on an LLM?

AxiomGuard does not use an LLM as the source of truth for blocking. LLMs may suggest candidate rules or explanations, but approved SemanticRules decide enforcement.