Drone Inspection Specialist Agent Skill
Advanced CV for infrastructure inspection including forest fire detection, wildfire precondition assessment, roof inspection, hail damage analysis, thermal imaging, and 3D Gaussian Splatting reconstruction. Expert in multi-modal detection, insurance risk modeling, and reinsurance data pipelines. Activate on "fire detection", "wildfire risk", "roof inspection", "hail damage", "thermal analysis", "Gaussian Splatting", "3DGS", "insurance inspection", "defensible space", "property assessment", "catastrophe modeling", "NDVI", "fuel load". NOT for general drone flight control, SLAM, path planning, or sensor fusion (use drone-cv-expert), GPU shader development (use metal-shader-expert), or generic object detection without inspection context (use clip-aware-embeddings).
Install
one command, takes just this skill from the repositorynpx skills add https://github.com/curiositech/some_claude_skills --skill drone-inspection-specialistThe instruction itself
26 sections, as written by the authorDrone Inspection Specialist
Expert in drone-based infrastructure inspection with computer vision, thermal analysis, and 3D reconstruction for insurance, property assessment, and environmental monitoring.
Decision Tree: When to Use This Skill
User mentions drones/UAV?
├─ YES → Is it about inspection or assessment of something?
│ ├─ Fire detection, smoke, thermal hotspots → THIS SKILL
│ ├─ Roof damage, hail, shingles → THIS SKILL
│ ├─ Property/insurance assessment → THIS SKILL
│ ├─ 3D reconstruction for measurement → THIS SKILL
│ ├─ Wildfire risk, defensible space → THIS SKILL
│ └─ NO (flight control, navigation, general CV) → drone-cv-expert
└─ NO → Is it about fire/roof/property assessment without drones?
├─ YES → Still use THIS SKILL (methods apply)
└─ NO → Different skill needed
Core Competencies
Fire Detection & Wildfire Risk
- Multi-Modal Detection: RGB smoke + thermal hotspot fusion
- Precondition Assessment: NDVI, fuel load, vegetation density
- Defensible Space: CAL FIRE/NFPA 1144 compliance evaluation
- Progression Tracking: Spread rate, direction prediction
Roof & Structural Inspection
- Damage Detection: Cracks, missing shingles, wear, ponding
- Hail Analysis: Impact pattern recognition, size estimation
- Thermal Analysis: Moisture detection, insulation gaps, HVAC leaks
- Material Classification: Asphalt, metal, tile, slate identification
3D Reconstruction (Gaussian Splatting)
- Pipeline: Video → COLMAP SfM → 3DGS training → Web viewer
- Measurements: Roof area, damage dimensions, property bounds
- Change Detection: Before/after comparison for claims
Insurance & Reinsurance
- Claim Packaging: Documentation meeting industry standards
- Risk Modeling: Catastrophe models, loss distributions
- Precondition Data: Satellite + drone + ground integration
Anti-Patterns to Avoid
1. "Single-Sensor Dependence"
Wrong: Using only RGB for fire detection.
Right: Multi-modal fusion (RGB + thermal) for high-confidence alerts.
| Detection Source | Confidence | Action |
|------------------|------------|--------|
| Thermal fire only | 70% | Alert + verify |
| RGB smoke only | 60% | Alert + investigate |
| Thermal + RGB | 95% | Confirmed fire |
2. "Ignoring Hail Pattern"
Wrong: Counting damage without analyzing spatial distribution.
Right: True hail damage has RANDOM distribution. Linear or clustered patterns indicate other causes (foot traffic, age).
3. "Thermal Temperature Trust"
Wrong: Using raw thermal values without calibration.
Right: Account for:
- Emissivity of materials (roof = 0.9-0.95)
- Atmospheric transmission (humidity, distance)
- Reflected temperature from surroundings
- Time of day (thermal lag)
4. "3DGS Frame Overload"
Wrong: Extracting every frame from drone video.
Right: Extract 2-3 fps with 80% overlap. More frames ≠ better reconstruction.
| Video FPS | Extract Rate | Result |
|-----------|--------------|--------|
| 30 | 30 (all) | Redundant, slow processing |
| 30 | 2-3 | Optimal quality/speed |
| 30 | 0.5 | Insufficient overlap |
5. "Insurance Claim Speculation"
Wrong: Estimating costs without material identification.
Right: Identify material → Apply correct cost matrix.
| Material | Repair $/sqft | Replace $/sqft |
|----------|--------------|----------------|
| Asphalt shingle | $5-10 | $3-7 |
| Metal | $10-15 | $8-14 |
| Tile | $12-20 | $10-18 |
| Slate | $20-40 | $15-30 |
6. "Defensible Space Zone Confusion"
Wrong: Treating all vegetation equally regardless of distance.
Right: CAL FIRE zones have different requirements:
| Zone | Distance | Requirement |
|------|----------|-------------|
| 0 | 0-5 ft | Ember-resistant (no combustibles) |
| 1 | 5-30 ft | Lean, clean, green (spaced trees) |
| 2 | 30-100 ft | Reduced fuel (selective thinning) |
Data Collection Strategy
Satellite Data (Regional Context)
- Sentinel-2: 10m resolution, NDVI, fuel moisture (SWIR bands)
- Landsat-8: 30m resolution, historical baseline, thermal band
- Planet: 3m resolution daily, change detection
- Application: Regional risk mapping, before/after events
Drone Data (Property Detail)
- RGB Mapping: 2-5cm GSD, orthomosaic, 3D model
- Thermal Survey: Moisture detection, heat signatures
- Close Inspection: Damage documentation, detail photos
- Application: Individual property assessment
Ground Truth
- Slope Measurement: GPS transects for topographic risk
- Soil Sampling: Moisture content for fire risk
- Material Verification: Confirm roof type
- Application: Calibration and validation
Quick Reference Tables
Fire Detection Confidence Levels
| Signal Combination | Confidence | Alert Priority |
|-------------------|------------|----------------|
| Thermal >150°C + Smoke | 95% | CRITICAL |
| Thermal fire model | 80% | HIGH |
| Hotspot >80°C | 70% | MEDIUM |
| Smoke only | 60% | MEDIUM |
| Hotspot 60-80°C | 50% | LOW |
Roof Damage Severity
| Type | Low | Medium | High | Critical |
|------|-----|--------|------|----------|
| Missing shingle | - | - | Always | - |
| Crack | <1" | 1-3" | >3" | Multiple |
| Granule loss | <10% | 10-30% | >30% | - |
| Ponding | - | Small | Large | Active leak |
Wildfire Risk Factors (Weighted)
| Factor | Weight | High Risk Indicators |
|--------|--------|---------------------|
| Defensible space | 20% | Non-compliant zones |
| Vegetation density | 20% | NDVI >0.6, high fuel load |
| Slope | 15% | >30% grade |
| Roof material | 10% | Wood shake, Class C |
| Structure spacing | 10% | <30ft between buildings |
| Access/egress | 10% | Single road, narrow |
3DGS Quality Settings
| Quality Level | Iterations | Time | Use Case |
|---------------|------------|------|----------|
| Preview | 7K | 5 min | Quick check |
| Standard | 30K | 30 min | General use |
| High | 50K | 60 min | Documentation |
| Inspection | 100K | 3 hrs | Damage measurement |
Reference Files
Detailed implementations in references/:
fire-detection.md- Multi-modal fire detection, thermal cameras, progression trackingroof-inspection.md- Damage detection, thermal analysis, material classificationinsurance-risk-assessment.md- Hail damage, wildfire risk, catastrophe modeling, reinsurancegaussian-splatting-3d.md- COLMAP pipeline, 3DGS training, inspection measurements
Integration Points
- drone-cv-expert: Flight control, navigation, general CV algorithms
- metal-shader-expert: GPU-accelerated 3DGS rendering
- collage-layout-expert: Visual report composition
- clip-aware-embeddings: Material/damage classification assistance
Insurance Workflow
1. Pre-Event Assessment (Underwriting)
├─ Satellite: Regional risk context
├─ Drone: Property-level risk factors
└─ Output: Risk score, premium factors
2. Post-Event Inspection (Claims)
├─ Drone survey: Damage documentation
├─ 3DGS: Measurements, change detection
└─ Output: Claim package, cost estimate
3. Portfolio Risk (Reinsurance)
├─ Aggregate: TIV, loss curves
├─ Model: AAL, PML, concentration
└─ Output: Treaty pricing, structure
Key Principle: Inspection accuracy depends on multi-source data fusion. Single-sensor assessments miss critical context. Always correlate drone findings with satellite baseline and weather data for defensible conclusions.
Other skills for the same job
different authors, same section of the catalogue
Production-ready patterns for building LLM applications. Covers RAG pipelines, agent architectures, prompt IDEs, and LLMOps monitoring. Use when designing AI applications, implementing RAG, building agents, or setting up LLM observability.
Build production ML systems with PyTorch 2.x, TensorFlow, and modern ML frameworks. Implements model serving, feature engineering, A/B testing, and monitoring. Use PROACTIVELY for ML model deployment, inference optimization, or production ML infrastructure.
World-class ML engineering skill for productionizing ML models, MLOps, and building scalable ML systems. Expertise in PyTorch, TensorFlow, model deployment, feature stores, model monitoring, and ML infrastructure. Includes LLM integration, fine-tuning, RAG systems, and agentic AI. Use when deploying ML models, building ML platforms, implementing MLOps, or integrating LLMs into production systems.
Expert in Langfuse - the open-source LLM observability platform. Covers tracing, prompt management, evaluation, datasets, and integration with LangChain, LlamaIndex, and OpenAI. Essential for debugging, monitoring, and improving LLM applications in production. Use when: langfuse, llm observability, llm tracing, prompt management, llm evaluation.
Use this skill for reinforcement learning tasks including training RL agents (PPO, SAC, DQN, TD3, DDPG, A2C, etc.), creating custom Gym environments, implementing callbacks for monitoring and control, using vectorized environments for parallel training, and integrating with deep RL workflows. This skill should be used when users request RL algorithm implementation, agent training, environment design, or RL experimentation.
Managed vector database for production AI applications. Fully managed, auto-scaling, with hybrid search (dense + sparse), metadata filtering, and namespaces. Low latency (<100ms p95). Use for production RAG, recommendation systems, or semantic search at scale. Best for serverless, managed infrastructure.
Deploy, evaluate, fine-tune, and manage Foundry agents end-to-end with azd: hosted agent scaffold/run/deploy, prompt agent create, batch eval, continuous eval, prompt optimizer, Agent Optimizer scaffold, agent.yaml, dataset curation from traces, model fine-tuning (SFT/DPO/RFT). USE FOR: azd ai agent, azd provision/deploy, deploy agent, hosted agent, create agent, add tool to agent, invoke agent, evaluate agent, continuous eval, continuous monitoring, agent CI/CD, optimize prompt, improve prompt, optimize agent instructions, agent optimizer, deploy model, Foundry project, RBAC, role assignment, permissions, quota, capacity, region, troubleshoot agent, deployment failure, AI Services, create Foundry resource, provision, knowledge index, customize deployment, onboard, availability, fine-tune, SFT, DPO, RFT, training-data, grader, distillation, fine-tuned model, large file upload. DO NOT USE FOR: Azure Functions, App Service, general Azure deploy (use azure-deploy), general Azure prep (use azure-prepare).
Cost optimization patterns for LLM API usage — model routing by task complexity, budget tracking, retry logic, and prompt caching.
How to use it
Take curiositech/drone-inspection-specialist from the repository into ~/.claude/skills for personal
use, or into .claude/skills inside a project.
The agent identifies a skill by the name field in its header. Two skills with the
same name cannot sit side by side — one of them will be ignored.