code-review-criteria

Code review criteria are generated from ADRs by extracting architecture decisions and applying code-convention filters to produce module-specific baselines. This automated workflow uses isolated subagents to extract ADRs and merge them into a unified code-quality guideline.

Automating pre-merge code review guidelines involves mapping module-specific code-convention files against extracted ADRs. This produces reproducible evaluation baselines that ensure consistent quality checks before merging code across multiple modules.

ADR-driven evaluation baselines can be created independently for multiple modules. The skill supports a two-stage workflow that extracts ADRs and merges evaluation criteria separately for each targeted module path.

ADR extraction works with code convention filters by first pulling architectural decision records and then applying relevant conventions to format the output. This mapping creates reproducible, module-specific review criteria.

A code-convention.yaml file is needed to map against extracted ADRs and produce reproducible review criteria. You must provide the target module path and ADR location before invoking the extraction and generation workflow.

Automated code review criteria generation from ADRs is limited to producing baselines for pre-merge reviews and independent ADR-driven criteria creation. It requires accurate ADR paths and module-specific convention files to function correctly.