legal-sanity-review
À propos
Ce processus obligatoire analyse le code pour détecter les risques juridiques, tels que les noms de clients ou les références propriétaires, avant toute revue croisée. Il bloque le cycle de revue si des violations sont détectées, exigeant des corrections préalables. Utilisez-le comme pré-verrou pour la création de PR, le portage de code ou tout déclencheur de revue croisée.
Installation rapide
Claude Code
Recommandénpx skills add vamseeachanta/workspace-hub/plugin add https://github.com/vamseeachanta/workspace-hubgit clone https://github.com/vamseeachanta/workspace-hub.git ~/.claude/skills/legal-sanity-reviewCopiez et collez cette commande dans Claude Code pour installer cette compétence
Documentation
Legal Sanity Review Workflow
Version: 1.0.0 Category: Workflows Triggers: Before cross-review (Codex/Gemini), PR creation, code porting
Purpose
Ensures no client project names, proprietary tool references, or legally sensitive content reaches the cross-review stage or a pull request. This is a mandatory pre-gate — if the legal scan fails, the review cycle does not proceed.
Review Flow
Claude/Gemini performs task
↓
Commit changes
↓
Legal Sanity Scan ◄── MANDATORY PRE-GATE
├── BLOCK → Fix violations → Re-scan
└── PASS → Proceed to Codex review
↓
Cross-Review Cycle
Integration Points
Cross-Review Pre-Gate
The legal scan runs before the first Codex/Gemini iteration:
| Step | Gate | Action |
|---|---|---|
| 1 | Legal Scan | Run legal-sanity-scan.sh --diff-only |
| 2 | Pass? | If no → fix and re-scan |
| 3 | Codex Review | First cross-review iteration |
| 4 | Gemini Review | Parallel cross-review |
PR Pre-Hook
Integrated into pr-manager.md hooks:
./scripts/legal/legal-sanity-scan.sh --diff-only || (echo "Legal sanity FAILED" && exit 1)
Manual Invocation
# Full repo scan
./scripts/legal/legal-sanity-scan.sh --repo=worldenergydata
# All submodules
./scripts/legal/legal-sanity-scan.sh --all
Exit Conditions
| Condition | Code | Action |
|---|---|---|
| PASS | 0 | Proceed to cross-review |
| BLOCK | 1 | Fix violations, re-scan |
| SCAN_ERROR | 2 | Check script/config, retry |
Violation Response
When the scan finds block-severity violations:
- Stop — do not proceed to cross-review or PR creation
- Report — list each violation with file:line and pattern
- Fix — replace client references with generic equivalents
- Re-scan — confirm all violations are resolved
- Continue — proceed to the cross-review cycle
Configuration
Deny lists control what patterns are scanned:
- Global:
.legal-deny-list.yaml(workspace root) - Per-project:
<submodule>/.legal-deny-list.yaml
See .claude/rules/legal-compliance.md for the full policy.
Use this workflow whenever committing AI-generated or ported code to ensure legal compliance before review.
Dépôt GitHub
Compétences associées
brainstorming
AutreThe brainstorming skill transforms initial ideas into detailed specifications through iterative questioning and collaborative dialogue. It helps developers explore design options, evaluate trade-offs, and create shared understanding before implementation. This prevents rework by systematically refining concepts into actionable plans.
repo-cleanup-progress-tracking-commands
AutreThis sub-skill provides shell commands for tracking file counts during repository cleanup operations, specifically for hidden folders like `.claude` and `.agent-os`. It includes verification scripts to ensure no files are lost after merging changes. Use it when consolidating or reorganizing project directories to monitor progress and validate results.
brainstorming-metrics
AutreThis skill provides structured metrics for evaluating brainstorming sessions, including targets for questions per design, validation rates, and stakeholder alignment. It helps developers measure brainstorming effectiveness and implementation quality. Use it to establish clear success criteria and track rework during project planning phases.
brainstorming-error-handling
AutreThis Claude Skill provides structured guidance for handling common brainstorming roadblocks like conflicting requirements or scope creep. It offers specific action templates to resolve issues such as unclear responses or analysis paralysis. Developers should use this reference skill to maintain productive brainstorming sessions within Claude Code.
