--- name: workflow-compliance-supervisor description: Use when a UI/UX project must verify that the promised user journey, TASK-PLAN.md, progress-state.json, rendered progress HTML, evidence artifacts, and skipped stages are consistent. Detects lazy skips, fake Ran/Done claims, missing evidence, approval-gated external steps, and user-data blockers; recommends continue, pause_for_user, return_to_rework, skip_with_reason, or block without replacing senior-ui-ux-orchestrator as chair. --- # 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 validate `workflow-log.jsonl` events. - `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 1. Read [references/decision-framework.md](references/decision-framework.md). 2. Read [references/evidence-and-log-rules.md](references/evidence-and-log-rules.md). 3. Read [references/user-messaging.md](references/user-messaging.md) if the output will be shown to a non-technical user. 4. Locate the canonical project `TASK-PLAN.md`. 5. Locate the declared journey registry. Default local registry: `$UIUX_SKILL_SYSTEM_ROOT/data/journey-registry.json`. 6. Compare declared journey stages with task-plan claims, progress state, reports, screenshots, and logs. 7. Produce one decision per affected stage: `continue`, `pause_for_user`, `return_to_rework`, `skip_with_reason`, or `block`. 8. For every skip, require an acceptable reason code, impact, and next action. 9. For every `done`/`Ran` claim, require real evidence. 10. Write or update a sanitized report/log only when the user asked for execution or validation. ## Decision States - `continue`: evidence is sufficient or deviation has no material impact. - `pause_for_user`: missing user data, approval, budget, credentials, or production confirmation. - `return_to_rework`: claimed completion lacks evidence or violates the journey contract. - `skip_with_reason`: stage is intentionally skipped with accepted reason and visible impact. - `block`: continuing would compromise quality, privacy, legality, production safety, or journey integrity. ## Hard Rules - Do not accept `model_convenience`, `implicit_assumption`, `planned_counted_as_ran`, or `tool_not_called_but_marked_done`. - Do not mark planned commands as executed. - Do not claim that a full journey is complete when only a demo slice ran. - Do not store secrets, tokens, cookies, `.env`, recovery codes, private screenshots, or raw provider payloads. - Optional external tools, crawls, uploads, paid APIs, production deploys, analytics, ads, and account mutations require explicit approval. ## Local Validation Commands ```bash python3 "$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-supervisor ``` ## Handoff Shape ```text Declared journey: Checked artifacts: Decision: Reason code: Evidence: User-visible message: Next action: Can continue: Must rework: Skipped with accepted reason: Blocked: ```