Skip to main content

Run log

2026-06-28 — version 1​

  • Implemented a seeded 240-resident workforce pipeline.
  • Added controls for budget, outreach, concurrent training seats, and job openings.
  • Added ODD protocol, code notes, and first controlled-experiment tutorial.
  • Capture target: default model and high-capacity comparison in the real Docusaurus site.

Debug order: URL/base path → simulator selector → control selector → result render → screenshot path.

2026-06-28 — Mesa source and headed browser evidence​

  • Replaced the TypeScript simulation engine with Mesa 3.5.1 Model, Agent, and DataCollector output.
  • Generated four seeded scenario series in mesa-results.json.
  • Generated four seeded scenario run folders under runs/ and wrote runs/index.json plus mesa-results.json for compatibility.
  • Ran the production Docusaurus build through headed Chrome using Playwright.
  • Captured baseline, outreach 16%, training seats 72, and openings 160.
  • Persisted observed browser metrics in experiment-results.json.
  • Disabled headed-Chrome GPU occlusion and waited for Docusaurus hydration before capture.

Capture URL: http://127.0.0.1:3000/simulations/workforce-development/city-opportunity-simulator.

2026-06-28 — internal research cockpit​

  • Added a standalone Vite/React/Phaser dashboard that reads the checked-in Mesa run bundles.
  • Added whole-board replay, pan and zoom, pause and timeline controls, scenario reports, and default dark-mode AutoNateAI branding.
  • Added single-agent and Shift-click group following with persistent camera framing and visible follow markers.
  • Documented the roadmap for Codex-assisted board construction, homes, mobility modes, physics, research capture, and Docusaurus/PDF report production.
  • Captured the baseline dashboard at week 2 for the public roadmap and dashboard README.

Capture URL: http://127.0.0.1:5174/. Scenario: workforce_001-baseline-seed42. Asset: simulations/assets/sim-dashboard-dark.png.

2026-06-28 — hourly spatial city model​

  • Added a Mesa-authored city contract with a residential street grid, education organizations, support services, and a business district.
  • Distributed resident homes deterministically along streets and assigned individual weekday start/end schedules.
  • Added walking, biking, and driving modes with seeded speed variation and finite road-constrained travel.
  • Added deterministic lane separation as the first collision-avoidance rule; agents cannot teleport between destinations or travel across turf.
  • Added a 2,689-position hourly clock across 16 seven-day weeks and connected the Phaser replay to hour/day/week state.
  • Added dawn, daylight, dusk, and night overlays and updated the dashboard capture at week 0, day 1, 11:00.

Spatial artifacts: world.json and hourly_clock.csv.

2026-06-28 — continuous travel and time access​

  • Replaced direct coordinate tweens with continuous interpolation along Mesa-authored route segments.
  • Advanced the dashboard in 15-minute steps and treated start time as target arrival time, making commute distance visible.
  • Added directional lanes, signal stops, following gaps, queue delay, home entry, and hidden at-home state.
  • Added resident priority, active days, weekly time budget, action hours, commute hours, and time-access effects to every pipeline transition.
  • Regenerated all four seeded Mesa run bundles under model version 2.1.0.

Dashboard verification URL: http://127.0.0.1:5174/. Tutorial capture URL: http://127.0.0.1:3001/simulations/workforce-development/city-opportunity-simulator (headed Chrome). Final tutorial image files were re-rendered with the same script in headless mode after macOS headed-window occlusion produced corrupted pixels.

  • Added persistent destination labels for the support center, north and south training centers, and west and east employer hubs.