abd-build-practice-scanners

Automating validation for DO and DO NOT rules involves turning them into machine-checkable tests via scanner scripts that read rules/*.md. This enforces rule intent consistently and catches drift early during authoring and rollout.

The best way to enforce consistent quality rules is by bundling them into a stable SKILL.md and running automated scanners. This ensures consistent validation across large rule sets and provides fast feedback on drift.

To bundle written rules into a machine-checkable format, run the bundle_rules_into_skill_md.py script against your rules/*.md files. This updates SKILL.md and wires each rule to its corresponding scanner script.

Yes, you need a specific directory structure: a stable SKILL.md, a rules/*.md directory for the written rules, and a scanners/ directory containing the Python scripts that execute the automated validation checks.

Automated rule validation fails to catch drift if scanner wiring is incomplete. Each rule must specify a scanner in its frontmatter and have a corresponding script in the scanners directory to execute checks properly.

You should not use automated scanners for rule enforcement when your SKILL.md is unstable or your DO/DO NOT rules are not yet finalized. These scanners require a stable SKILL.md and are designed for the authoring and rollout phases.