Topic: Game-Theoretic AI for Robust COA Generation and Wargaming
Close: 2026-06-24 noon ET · Vehicle: AFWERX 26.BZ Release 2 / OSD SCO
DSIP: topic details
Capability truth: docs/canonical/11_r2_capability_truth.md · Prove: GET /api/prove/pursuit-r2
| Claim | Live API | Phase II hardening |
|---|---|---|
| Adversary-branch exploration | POST /api/wargame/match |
Deeper tree + M&S export (CPE/AFSIM) |
| Interpretable doctrinal phases | branchTree[].doctrinalPhase |
Tie to 131A cycle labels in UI |
| Quantified speed vs human baseline | GET /api/wargame/benchmark |
Witnessed red-team exercises |
| Human release on execution | targeting-131a + strike-ot approve pattern | Formal benchmark report |
Related live lanes: /api/targeting-131a/full-cycle, /api/coa/generate · Module: server/pursuit-r2-capabilities.ts
Only pursue if NV007 + NV006 are certified first. This topic accepts Direct to Phase II proposals only — not a standard $100K Phase I.
server/pursuit-r2-capabilities.ts — wargame match + benchmark APIsserver/targeting-131a-engine.ts — deliberate targeting cycleserver/coa-engine.ts — COA generation APIserver/fire-mission-generator.ts — fires planningDo not claim a Libratus-class Nash equilibrium solver. Position as doctrinally grounded adversary-branch exploration with human certification — aligned to topic's interpretability requirement.
Overlaps DARPA DISCORD (Jul 9). DV004 = SBIR D2P2; DISCORD = white-paper RFI. Do not copy-paste narratives.
This page works best with your current GPS location.