🎮 Sentinel Scout: Spatial Observability Engine

Sentinel Scout turns raw VALORANT positional telemetry into actionable scouting insights. It treats player movement as an event stream and applies unsupervised learning to infer default setups like defensive anchors and offensive entry formations.

🛡️ Technical Philosophy: "Resilient Observability"

• Constraint: Encountered a 403 Forbidden entitlement block on high-fidelity series-state telemetry via the GRID Open Access tier.
• Pivot: Architected a data-source-agnostic pipeline validated on schema-accurate mock telemetry. Crucially, clustering is computed at runtime from raw (x, y) coordinates—the engine is logic-complete and ready for production data.

🚀 Key Features

• Unsupervised Spatial Learning: Uses K-Means (K=2) to learn tactical zones directly from raw coordinate streams without pre-labeled data.
• Euclidean Spread Metric: Calculates the average Euclidean distance to centroids to quantify team discipline (“Tight Stack” vs “Distributed”).
• Automated Intelligence Artifacts: Generates a structured terminal report and a saved tactical zone plot (scouting_map.png).
• Headless Architecture: Uses the Matplotlib Agg backend for reliable execution in CI/CD and containerized environments.

⚡ Quickstart

# Install dependencies
pip install -r requirements.txt

# Execute the scouting engine
python3 scout_rigor.py
Outputs:

Terminal: Structured scouting report (centroids, zone share, spread, confidence).

File: scouting_map.png (Spatial distribution plot).

📊 Live Output (Example)
Plaintext
🏆 --- COMPETITION-GRADE SCOUTING REPORT (FINAL) ---
🤖 Model: K-Means Clustering (K=2)
📏 Metric: Spread = Avg Euclidean distance to team centroid

📍 Opponent (DEF Defaults):
   - Position: Backsite/Anchor (Centroid ≈ [-490, 2125])
   - Spread Metric: 26.93 (Tight Stack)
   - Confidence: Low (2 samples)

📍 Cloud9 (ATT Defaults):
   - Position: Mid/Entry (Centroid ≈ [1050, 4388])
   - Spread Metric: 277.59 (Distributed)
   - Confidence: Low (4 samples)
🔭 Roadmap: The Observability Giant
Coordinate Normalization: Scale logic across all VALORANT maps.

Dynamic K-Selection: Implement "Elbow Method" to auto-detect the number of active tactical zones.

Least Privilege: Hardening the execution environment in alignment with man7/capabilities.

Developed for the VCT Hackathon 2026 - Focus on Data Resilience and Spatial Analytics.