A native Windows background tool that scrapes AoE2:DE resource/villager counts from the screen in real time and combines them with replay data to produce post-game efficiency coaching. Target: minimal CPU footprint (<1% to avoid impacting the game), no memory hooking, no anti-cheat risk.
- Python R&D (
research/) is finished and remains the canonical reference. - The Rust production binary is fully functional, with a complete vision pipeline operating under 1% CPU overhead.
- Master Orchestrator is operational, automating capture, replay discovery, data meshing, and analysis.
- Interactive, offline-capable HTML reports are automatically generated and bundled with Chart.js for standalone use.
aoe2-squire/
├── CLAUDE.md ← you are here
├── DECISION_LOG.md ← all architectural decisions with rationale — read this
├── TECHNICAL_SPEC.md ← product-level requirements
├── Cargo.toml ← Rust manifest
├── ui_map.json ← element coordinates (baseline 1080p, scaled at runtime)
├── src/
│ ├── main.rs ← entry point
│ ├── lib.rs ← module definitions
│ ├── constants.rs ← all pipeline constants
│ ├── analysis/ ← modular analyzer implementation
│ ├── capture/ ← DXGI screen capture implementation
│ ├── pipeline/ ← Vision pipeline implementation
│ ├── bin/ ← Standalone CLI tools
│ ├── replay.rs ← replay parsing logic
│ ├── replay_discovery.rs ← automated replay location
│ ├── report.rs ← HTML report generator
│ ├── sync.rs ← temporal calibration/sync
│ └── types.rs ← shared data structures
├── templates/ ← HTML report templates
├── test_bench/ ← reference screenshots
└── research/ ← Python R&D only (reference)