Mode d’emploi
- Copiez la commande et examinez le paquet avant de l’installer dans votre environnement agent.
- Consultez le document original pour les outils, identifiants et environnements requis.
- Décrivez les entrées et le livrable souhaité ; comparez la sortie aux sources.
npx skills add https://github.com/sergekostenchuk/ui-ux-agent-skill-system --skill workflow-compliance-supervisorAvant de commencer
- L’installation n’inclut ni abonnement au modèle, ni quota API, ni accès aux fichiers.
- Étoiles et installations reflètent l’activité, pas un test indépendant de qualité ou de sécurité.
Document Skill original
Rendu depuis le dépôt source en conservant titres, exemples, code, tableaux, liens et images.
Workflow Compliance Supervisor
This skill is the governance layer for journey-driven UI/UX projects. It checks whether the system did what it promised, whether skipped stages are justified, and whether the user sees honest reasons and next actions.
It does not execute design/build tasks and does not override senior-ui-ux-orchestrator.
Operating Modes
gate: check the current project before continuing or closing a phase.review: audit an existing task plan, progress state, progress screen, and evidence set.log: append or validateworkflow-log.jsonlevents.fixture-test: run lazy-skip and fake-evidence fixtures.explain: explain to the user why a stage is paused, skipped, returned to rework, or blocked.
Required Workflow
- Read references/decision-framework.md.
- Read references/evidence-and-log-rules.md.
- Read references/user-messaging.md if the output will be shown to a non-technical user.
- Locate the canonical project
TASK-PLAN.md. - Locate the declared journey registry. Default local registry:
$UIUX_SKILL_SYSTEM_ROOT/data/journey-registry.json. - Compare declared journey stages with task-plan claims, progress state, reports, screenshots, and logs.
- Produce one decision per affected stage:
continue,pause_for_user,return_to_rework,skip_with_reason, orblock. - For every skip, require an acceptable reason code, impact, and next action.
- For every
done/Ranclaim, require real evidence. - Write or update a sanitized report/log only when the user asked for execution or validation.
Decision States
continue: evidence is sufficient or deviation has no material impact.pause_for_user: missing user data, approval, budget, credentials, or production confirmation.return_to_rework: claimed completion lacks evidence or violates the journey contract.skip_with_reason: stage is intentionally skipped with accepted reason and visible impact.block: continuing would compromise quality, privacy, legality, production safety, or journey integrity.
Hard Rules
- Do not accept
model_convenience,implicit_assumption,planned_counted_as_ran, ortool_not_called_but_marked_done. - Do not mark planned commands as executed.
- Do not claim that a full journey is complete when only a demo slice ran.
- Do not store secrets, tokens, cookies,
.env, recovery codes, private screenshots, or raw provider payloads. - Optional external tools, crawls, uploads, paid APIs, production deploys, analytics, ads, and account mutations require explicit approval.
Local Validation Commands
bash
python3 "$UIUX_SKILL_SYSTEM_ROOT/scripts/validate_workflow_compliance.py" --fixtures
python3 "$UIUX_SKILL_SYSTEM_ROOT/scripts/validate_workflow_log.py" --fixtures
python3 "$UIUX_SKILL_SYSTEM_ROOT/scripts/validate_workflow_compliance.py" --project "$PROJECT_ROOT" --journey new-premium-site --out "$PROJECT_ROOT/reports/workflow-compliance-report.md" --log "$PROJECT_ROOT/workflow-log.jsonl"Skill package checks:
bash
python3 $CODEX_HOME/skills/.system/skill-creator/scripts/quick_validate.py $CODEX_HOME/skills/workflow-compliance-supervisor
python3 $CODEX_HOME/skills/senior-skill-architect/scripts/lint_production_skill.py $CODEX_HOME/skills/workflow-compliance-supervisorHandoff Shape
text
Declared journey:
Checked artifacts:
Decision:
Reason code:
Evidence:
User-visible message:
Next action:
Can continue:
Must rework:
Skipped with accepted reason:
Blocked:
