governance-validator

schema	fcvw/skill@1
name	governance-validator
description	Validate FCVW structure, schemas, ownership, states, and clean-template boundaries.
version	1.2.0
trigger_keywords	["validate governance","verify filesystem","plan state","validar governança","context map","reading trigger","orphan policy","integridade do framework"]
session_types	["audit","release","framework_upgrade"]

Governance validator

Purpose

Evaluate the human-readable FCVW invariants. The optional script automates deterministic checks; this skill owns interpretation and remediation decisions.

Profiles

• clean-template: project-profile placeholders are allowed; application histories and contamination are forbidden.
• instantiated: required project profiles must be complete.
• incremental: new violations fail; exact legacy baseline findings are reported but do not block.
• strict: all applicable findings block.

Inputs

• SCHEMAS.md, OWNERSHIP.md, FRAMEWORK_LOCK.md;
• physical filesystem;
• active plan and target release;
• optional legacy baseline.

Checks

1. Canonical files and ownership metadata.
2. Plan status/directory coherence and unique IDs.
3. Skill metadata, catalog coverage, and provider-neutral core instructions.
4. Framework/application version namespace separation.
5. Markdown links and fence balance.
6. Wiki IDs, schema, index, freshness, and archive rules.
7. Placeholder policy by profile.
8. Clean-template contamination.
9. Framework lock, release record, and migration coherence.
10. FILESYSTEM.md canonical path/glob coverage.
11. Operational-index and reading-route coverage for every root framework policy, project profile, and declared skill session type.

Optional execution

python tools/validate_fcvw.py --root . --profile <profile>

For controlled legacy debt:

python tools/validate_fcvw.py --root . --profile incremental --baseline path/to/legacy-baseline.md

Scripts do not auto-correct historical evidence. Review every proposed remediation and keep the report attached to the active plan or release.

Non-responsibilities

• application runtime tests;
• legal approval;
• destructive cleanup without a plan;
• hiding legacy findings.

Required output

List command/profile, passed checks, new failures, reading-route gaps, baseline findings, corrections, and residual risk.

Validation and exit

Exit only when blocking findings are resolved or explicitly accepted by authorized scope, and no new debt is suppressed as legacy.