| name | gomboc_community_flow_review_rule |
| description | Use when evaluating a rule package against release standards, auto-fixing blocking gaps, and producing a report. Depends on: gomboc_community_know_orl_runtime_resolution, gomboc_community_know_release_checklist, gomboc_community_cap_run_orl_test, gomboc_community_task_enrich_rule. |
Flow: Review Rule
Release-readiness engineer for ORL packages: evaluate gaps, auto-fix blocking issues and required metadata, ask only when values cannot be inferred. Checklist criteria live in gomboc_community_know_release_checklist — load that; do not restate the full checklist here.
Inputs
- Path to a rule package (
.orl, test.orl, workspace/, workspace_expected/)
Process
Step 1 — Test gate
gomboc_community_cap_run_orl_test. On failure, stop and report — do not auto-cleanup past failing tests.
Step 2 — Locate or create assessment
Read outputs/rule-assessment.md if present. Otherwise perform a lightweight 9-criterion assessment (completeness, accuracy, test coverage, maintainability, conciseness, side effects, tool usage, safe execution, metadata) and write outputs/rule-assessment.md.
Step 3 — Fix release-process gaps
Against gomboc_community_know_release_checklist, auto-fix when deterministic; ask when not.
| Area | Action |
|---|
| Naming | metadata.name = gomboc-ai/<lang>/<framework>/<requirement>; rename .orl to match (/ → -); update test.orl refs |
| Metadata | Infer display_name, description, provider/resource, scores when possible; default visibility/public-rule-bodies; ask for contributed-by / reviewer name |
| Prefer | gomboc_community_task_enrich_rule when filling standard community metadata fields |
test.orl | Add schema header; replace comparison: ast → mode: ast; fill expected counts from Step 1 |
| Workspace | Report file-count mismatches; strip comments from fixtures then re-test |
| Classifications | Prepend gomboc-ai/policy/ when missing |
| Outputs | Generate missing outputs/*.md placeholders (plan/build/review/test-plan/enrich/scanner-validation) |
Step 4 — Ask for missing values
Single consolidated prompt for anything that could not be inferred; apply answers.
Step 5 — Re-test
gomboc_community_cap_run_orl_test. On failure, revert the last risky change and report.
Step 6 — Re-assess
Archive prior assessment to outputs/rule-assessment-pre-cleanup.md. Produce a fresh outputs/rule-assessment.md reflecting post-fix state. Compare scores; if still DENY/CONDITIONAL, list remaining manual work.
Step 7 — Report
Write outputs/cleanup-report.md (changes made, re-assessment, recommended non-blocking improvements) and print it.
Ask vs auto-fix
| Situation | Action |
|---|
| Deterministic from rule content | Auto-fix |
| Safe default (e.g. visibility public) | Auto-fix with default |
| Reasonable inference | Auto-fix; note in report |
| Personal/org (contributor, reviewer) | Ask |
| Ambiguous / misrepresentation risk | Ask |
| Tests fail after a fix | Revert and report |
Out of scope
- New test cases / rewriting audit or remediation logic
- Push/release — use
gomboc_community_task_release_rule
