c149. Simulation Mode, Both Seats

Opened 2026-11-13 at c144's close. Mark was asked whether c144's unfinished phases should open a successor now, and answered "Open stage 1 now". c144 (Theatre Mode, now Simulation mode) built Phases 0, 1 and 5 and S1 in the native app, and closed with the rest carried forward. That rest is this cycle.

What it's about

c144's plan, from Phase 2 on (../c144.theatre-mode/plan.md § 9), with its rulings C1–C10 still standing:

• Both seats (Phase 2). Each scenario in both seats: the operator's fleet pages, and the person's laptop in the native skin. The walk at Observe and at Enforce on every surface. A harness (Claude Code) plays S1 from one prompt, through ?theatreTest=1 and the guide.
• The simulation's own messages (C6). Real messages arrive as usual. The simulation's own are marked ("Simulation · …") in the tray, the inbox and the OS banner. They live in the simulation's copy and leave with it. c137 closed into c147, so this now means c147's Findings and c4's plumbing, not an inbox.
• S2–S6 in the native app (Phase 6). Each scenario's beats go into the nozzle's synthetic source, the way S1's do (nozzle/src/theatre.rs). Their scenes are exported (skill-update, share-request, public-link, ticket-card, stranger-asks). c133's four new trainings (T-PROVIDERS-1, T-PROVENANCE-1, T-APPROVAL-1, T-SPEND-1, design 463237c3a1) are bundled with them, as T-INJECT-1 went with S1 (qnative c9).
• New here at three registers (Phase 3). I'm new to this, I work with computers, I have networking degrees, all written (C4). The native New to Qpoint tour exists at one register.
• AI, on your own machine (Phase 4). this-machine.json grows into the primer on real data: the metaphor (model as brain, harness as agent, tools as hands), where it breaks, and today's marks only (C5). Native first (C3).
• Review and port (Phase 7). What goes to q-nuxt-layer (the simulation panel, the register control, the tour look), what goes to qdash, and what stays lab-only.

Starts from

• c144's record: CYCLE.md (Outcome: what was made, what the play taught, the lessons), scenarios.md (§ As built, the gaps sent to c126) and the story at /cycles/design/c144.
• The lab:
  • the scenarios in app/data/scenarios/;
  • the simulation in useTheatre, components/theatre/ and data/tours/theatre.ts;
  • the checks in scripts/theatre-check.mjs (190/0 at c144's close).
• qnative: native S1 (d94716f), its stories tests, and the single-instance guard.
• Mark's rulings at c144's close (2026-11-13):
  • Priya is on Cursor CLI in the lab and in the native world;
  • the temporary look controls stay behind ?lookControls=1;
  • the native Models and Tools callouts hiding under the tour panel at 1120 px are qnative c12 Punch List item 2.

Ruled before kick-off: prevented rows in the simulation

c147's Q4 (Mark at its review stop 2: "c121 D3 superseded everywhere") resolves an Enforced finding at the moment it's detected. The simulation's world goes through the same fold, so enforced beat rows left Findings' Open list. With c147's uncommitted change in the tree on 2026-11-13, theatre-check was 184/6 and tour-drive-check 186/2.

Mark ruled at c147 review stop 3, call 1 (../c147.findings-common-tongue/review-3.md § Answered): during a simulation, Findings shows the prevented rows beside the open ones, marked Prevented. The beats' test lines and the tour's callouts expect them there. Q4 itself stays in the fold. This is c149's first build, after c147 commits Q4:

• Findings during a simulation shows prevented rows, marked;
• the scenarios' expect lines say so;
• theatre-check (the six: S1 and S4's replays, S2's and S5's sanitization at two beats each) and tour-drive-check's findings.group P-MCP check follow.

Two more rulings from the same stop touch the lab's model:

• a finished training that a policy assigned now closes its finding (this revises c133 D9 and isn't in foldFinding yet);
• chatgpt.com stays a separate provider from openai.com.

Not in it

• The policy display coloured by the finding's status (c144's follow-on, its own cycle when Mark picks it up).
• Wave 2's thirteen scenarios (scenarios.md), unless Mark pulls some in.
• New policies: the gaps stay with c126.

Stage 1 only: no CYCLE.md and no plan until Mark kicks it off.