Static vs semantic rules
Separate static rules (lint, types) from semantic invariants so you can enable modules gradually in large repos.
Case topic · extended category
Single-page topic Aligns with R&D & quality band
Overview: before merge and release, have agents run review checklists, test-gap analysis, and security-sensitive hints against your team baselines. Works with protected branches, required reviews, and CI gates: the skill can state what must be human-confirmed vs pre-filled.
Relationship to the case hub: the closest official category is R&D & quality (reviews, tests, licenses, and more). This page is a supplemental explainer to pair with those five cases; anchor #cat-dev.
Separate static rules (lint, types) from semantic invariants so you can enable modules gradually in large repos.
CI: align agent output with failing logs so you know if it’s policy or implementation. Tests: when coverage is missing, ask for minimal additive tests—not generic advice.
For legacy code or mixed stacks, ask for per-module policy in the skill so one rule set doesn’t misfire across the repo. Tag security-sensitive changes for stricter review lists.
# Case notes
- Trigger: PR opened / label needs-review
- Output: blockers + suggested tests
- Exit: blockers cleared or waived
Case hub · R&D & quality Open dev category index Cases overview