Reproduce stale upstream configuration with real local Git repositories, including the already-matching pin fast path.
Co-authored-by: Codex <codex@openai.com>
Record actual save field positions, motor phases, finite typed delivery, last-known radar contacts and blast-cover policy. Include per-seed baseline measurements and distinguish native/WASM checks from pending GUI acceptance.
Co-authored-by: Codex <codex@openai.com>
Use a bounded read-only pre-destruction occupancy sample; preserve conservative damage for unresolved geometry and do not defer or repeat damage on replies.
Co-authored-by: Codex <codex@openai.com>
Separate search and engagement roles, support launcher-only SAMs, animate individual tube lids and delay ground strike spending until ejection. Preserve battery configuration and launch poses in saves and replay.
Co-authored-by: Codex <codex@openai.com>
Share read-only scene data for smoke and displayed blasts; preserve cold ejection and flight-relative recording samples. Keep finite typed depot inventory, expose per-world smoke readiness through the pinned API26 engine, and invalidate presentation resources on world changes.
Validation: core suite 156 unit tests plus 4 scenarios and the new targeted recording regression; adapter 50 unit tests, boundary check and 4 scenarios; real WASM contract 16 steps; asset pipeline and bounds checks; macOS Metal client build and HD frame.
Co-Authored-By: Codex <noreply@openai.com>
Preserve launcher grids and authored materials; fix the Buk radar intersection and regenerate geometry bounds. Register the separate RM48U game visual for the upcoming carrier path.
Co-Authored-By: Codex <noreply@openai.com>
Keep authored JSON separate from generated output. Export rotated model bounds for SDK-free navigation and check grazing voxel corners. Include unretouched in-game before/after captures and explicit art/performance limits; pilots are an intermediate pass, not completion of the full asset set.
Co-Authored-By: Codex <noreply@openai.com>
Keep runtime compilation in an opt-in CI tier; run asset consistency on regular builds. Sandbox checks do not substitute for GPU, capture or maximum-scenario budgets.
Co-Authored-By: Codex <noreply@openai.com>
No compatible coordinate-strike loadout is implemented yet. Preserve flight and collision mechanics through explicit test fixtures without allowing generic stock or player position to create a production missile. Physical carriers, typed ammunition and logistics remain to implement.
Co-Authored-By: Codex <noreply@openai.com>
Capture actual missile launch/ignition, track transitions, fire state, SAM health and components. Keep legacy fallback and bound scene transitions to 65536. Renderers still use incremental adapters; shared SceneTimeline remains follow-up work.
Co-Authored-By: Codex <noreply@openai.com>
Reject narrow corridors, recheck motion clearance, and wait for footprint data before flooding the shared query budget.
Co-Authored-By: Codex GPT-6 Luna <noreply@openai.com>
Preserve the adapter facade and move adapter-only checks out of domain tests. Late tracks do not reveal an unseen launch origin or its timestamp.
Co-Authored-By: Codex GPT-6.1 Terra <noreply@openai.com>
Replace patch-stack reconstruction with safe pinned initialization and legacy checkout migration; use an engine-local packer and write artifact identities.
Co-Authored-By: Codex <noreply@openai.com>
In a clean real game with 15 systems: interceptors scrubbed at relief 58 to 2, kills 33 to 46 of 82. Smart raids from 2500 blocks: crashes 10 to 3.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Far raids over mountains with a 300-block known disc: crashes from 19 to 2 of 368 (mixed), 22 to about 9 (smart). Balance tests gain hilly land.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
chexit