game-systems-sandbox
Monte Carlo battle/balance analysis as a developer utility — balance your RPG combat with simulation.
A hybrid developer tool: deterministic TypeScript simulators, a Python/FastAPI bridge and compatibility layer, a shipped Next.js dashboard, and canonical offline evidence.
Why it matters
Seeded combat, loot, XP, economy, and A/B balance simulations make design changes measurable and reproducible. The dashboard runs against the API or deterministic browser-local demo data.
Verified evidence
14 focused release-contract tests; 173 Python tests with real-TSX integrations executed; 25 evidence tests; 56 root Vitest tests; 39 dashboard Vitest tests. Canonical evidence hash: c64e62c994476ffc2402fa4405f08239bbbcff8a18f7a3228b9c4b1436d4e932.
Offline by default
Local SQLite, an offline in-memory task registry, disabled telemetry, and dashboard demo data require no credentials or external service. PostgreSQL, Redis workers, and the best-effort HTTP telemetry adapter are optional. Browser demo mode visibly caps each request at 5,000 Monte Carlo fights or 200,000 loot rolls. Docker, the post-port Chromium final status, and a later full aggregate are not claimed green locally.
Deferred shelf scope
Parallel worker-thread/WASM execution, multi-fighter and status-effect combat, an agent economy, hosted experiments, auth, mandatory infrastructure, and storefront work remain deliberate future product boundaries.
Absorbed lineage
The package-private compatibility runtime preserves the proven MIT-licensed runtime module set absorbed from archived operator-shared-core v1.3.0 at source commit dbf276a7708da65b55e1f10b35af634b300d1f07. docparse.py is intentionally retained as optional transitive capability beyond the isolated runtime import proof; selected files were compared with the source commit and local patches are documented.
Repo facts
Default branch: main. Stars: 0. Forks: 0. Created: 2026-06-09.