ffca-audit
Audit an FFCA monorepo for architecture compliance: mechanical layer, naming, and cycle checks plus a qualitative source review, producing a per-package verdict table. Use in an FFCA monorepo, a featu
SKILL.mdskills/ffca-audit/SKILL.md
GitHubFFCA Audit
Whole-repo architecture health check. The hook keeps individual pubspec edits compliant; this skill answers whether the repo as a whole is healthy.
The audit itself is performed by the ffca-layer-auditor agent, so the work runs in its own context and can be reused by build, review, and pre-PR flows. This skill is the entry point that dispatches it.
What to do
- Confirm the repo is FFCA-shaped (a
features/folder with{feature}_{layer}packages). If not, say so and stop. - Launch the
ffca-layer-auditoragent (subagent typeffca-layer-auditor) against the repo. The agent runsdart run scripts/validate_layers.dart --allfor the mechanical pass (naming, layer rules, cycles), then adds the source-level checks (declared-but-unused dependencies, barrel hygiene, DTO leakage, Command/Query necessity, module entry, misplaced packages, high fan-in) and reads${CLAUDE_PLUGIN_ROOT}/references/ffca/for the detail behind each. - Relay the agent's output: the per-package verdict table and the severity-ordered findings list. Each finding states the rule and the fix. The agent is read-only and does not auto-fix.
If the agent is unavailable for any reason, fall back to running dart run ${CLAUDE_PLUGIN_ROOT}/scripts/validate_layers.dart --all yourself and reporting its violations, then note that the qualitative source-level checks were skipped.