fix(engine): health: field on pre-placed actors + building actor ids surfaced for repair/sell 22a6004 yxc20098 commited on May 21
feat(bench): then composite predicate happened-before (PlanBench replanning / PERT anchor) a5a3a5a yxc20098 commited on May 20
feat(bench): forbidden_tools + tool_violations_gte — strict-toolban / procedural-compliance primitive (BFCL V4 / τ²-bench / IFBench anchor) e3e91b7 yxc20098 commited on May 20
strategy-twobody: no-cheat redesign — two enemy bases + region-scoped destroy predicate; simultaneous two-group control is now load-bearing 2f8e4e6 yxc20098 commited on May 19
#2: enforce ordered waypoints (stateful waypoint_sequence) + two-path medium + scouted hard b165246 yxc20098 commited on May 19
Wire bench to vendored training prompt v2 (system/briefing/minimap) 8e88074 yxc20098 commited on May 19
Strategy packs: faithful 'destroy key economic buildings' objective e6d690e yxc20098 commited on May 18
Bench: surface S9 spatial tensor in render_state (multimodal reach) 41a0d2e yxc20098 commited on May 18
Catalog C13: harvest-economy packs — closes #14 (user economy families) 7a25eb3 yxc20098 commited on May 17
S1 bench wiring: resources/capacity + economy_value win predicates 1481b7f yxc20098 commited on May 17
adapter: use engine map_info for true map dims (S9), synthesis fallback 1b62e34 yxc20098 commited on May 17
Bench: consume S9 economy obs + economy win-conditions + full toolset 5a1cf72 yxc20098 commited on May 17
Add Rust-backed eval stack: scenario packs, adapter, spine, integration tests 098c6e0 yxc20098 commited on May 17