| name | opl-console-operator-copilot |
| description | Use when interpreting OPL Console/App operator projections, current_owner_delta, action catalogs, first-run readiness wording, Settings IA, Runtime page task awareness, user workbench actions, release evidence refs, and forbidden claims without executing actions or claiming App, domain, runtime, release, or readiness status. |
OPL Console Operator Copilot
Boundary
- Treat Console/App projections, Settings Control Center, first-run/onboarding refs, Runtime page projections, user workbench action catalogs, release evidence refs, current_owner_delta, runtime/readback surfaces, and domain owners as the authorities for their own facts.
- Use this skill only to interpret current projections, compare legal next actions, review operator-facing wording, and explain forbidden claims to an operator.
- Do not execute actions, enqueue work, mutate runtime state, sign owner receipts, create typed blockers, change domain truth, assert artifact authority, or make App release/domain/runtime readiness claims.
- If projection evidence and owner/runtime evidence disagree, classify the mismatch and route to the owning surface instead of choosing a truth source.
Workflow
- Identify the operator question:
current_owner_delta, action_catalog, projection_readback, next_action_choice, first_run_or_onboarding, settings_ia, runtime_task_awareness, user_workbench_action, release_evidence_ref, forbidden_claim_review, or route_back.
- Read the current projection refs and any linked owner/runtime refs. Do not infer beyond the refs provided.
- Classify available actions:
safe_read: inspect or explain state;
candidate_prepare: prepare a packet or evidence brief;
owner_route: requires domain, runtime, App, or human owner action;
forbidden: would mutate authority, execute an action, or overclaim readiness.
- For each plausible next action, state owner, required input, expected output ref, forbidden side effects, and proof needed after execution by the real owner.
- For App/operator UX questions, separate
Core safe to enter, Full/background pending, secondary/deep-link/advanced route, release evidence candidate, and not proven so ordinary navigation is not blocked by deep diagnostics or unproved release claims.
- Prefer the smallest legal operator move. If no legal move exists from the current projection, produce a route-back or missing-evidence note.
- Keep wording claim-safe: distinguish
projection says, owner ref proves, runtime readback proves, and not proven.
Legacy Coverage
This consolidated copilot covers the retired opl-app-first-run-ux-reviewer, opl-app-release-evidence-reviewer, opl-app-settings-ia-reviewer, opl-runtime-task-awareness-reviewer, and opl-user-workbench-action-reviewer entries. Keep those App/Console operator concerns here instead of restoring separate default metadata entries.
Output Shape
Return:
operator_question;
projection_refs;
owner_or_runtime_refs;
current_interpretation;
action_options: grouped as safe_read, candidate_prepare, owner_route, and forbidden;
operator_ux_review: first-run, Settings IA, Runtime page, workbench action, or release evidence wording when relevant;
recommended_next_action;
proof_needed_after_owner_action;
forbidden_claims: no action execution, no owner receipts, no typed blockers, no domain truth, no artifact authority, no App release/domain/runtime readiness claims.