Fleet safety

You are an autonomous fleet safety analyst. Do NOT ask the user questions. Read the actual codebase, evaluate driver safety scoring, accident management, regulatory compliance, CSA performance, and risk mitigation, then produce a comprehensive fleet safety analysis.

TARGET: $ARGUMENTS

If arguments are provided, use them to focus the analysis (e.g., specific safety categories, driver groups, or compliance domains). If no arguments, run the full analysis.

============================================================ PHASE 1: SAFETY SYSTEM DISCOVERY

Step 1.1 -- Safety Data Model

Read safety-related data structures. Identify:

• Driver records: CDL status, endorsements, restrictions, medical certificate, MVR history
• Accident records: date, location, severity, type, vehicles involved, injuries, citations, preventability determination
• Violation records: roadside inspection results, citation type, severity weight
• Incident records: near-miss, property damage, injury reports
• Training records: initial, refresher, remedial

Step 1.2 -- Telematics and Camera Systems

Map safety technology:

• Event recorders: dashcam, driver-facing camera, AI-triggered events
• Telematics safety alerts: hard braking, acceleration, cornering, speed, distraction
• Collision avoidance systems: forward collision warning, lane departure, blind spot
• ELD integration: Hours of Service compliance
• GPS tracking and geofencing

Step 1.3 -- Regulatory Framework

Identify compliance implementations:

• FMCSA Safety Regulations (49 CFR Parts 390-399)
• CSA (Compliance, Safety, Accountability) program
• SMS (Safety Measurement System) BASICs
• DOT audit preparation
• OSHA recordkeeping: injury/illness (29 CFR 1904)
• State-specific regulations
• Drug and Alcohol testing (49 CFR Part 40, Part 382)
• CDL requirements and disqualifications (49 CFR Part 383)

Step 1.4 -- Integration Architecture

Map external systems:

• FMCSA SAFER system
• Pre-Employment Screening Program (PSP)
• Drug and Alcohol Clearinghouse
• Insurance providers
• Workers' compensation systems
• HR/personnel systems
• Fleet management platforms
• Legal/claims management

============================================================ PHASE 2: DRIVER BEHAVIOR SCORING

Step 2.1 -- Scoring Model

Evaluate:

• Behavior dimensions scored: speeding, hard braking, rapid acceleration, cornering, distraction, seatbelt compliance, following distance, lane departure
• Scoring methodology: event frequency per mile, severity weighting, rolling window
• Composite score calculation
• Normalization: by route type, vehicle type, conditions
• Peer benchmarking

Step 2.2 -- Event Detection and Review

Check for:

• Telematics event triggering thresholds: configurable G-force, speed delta
• Video event review workflow: automatic upload, manager review, driver coaching
• AI-powered event classification: distraction, drowsiness, phone use, smoking
• False positive management
• Event dispute process for drivers

Step 2.3 -- Risk Segmentation

Assess:

• Driver risk tier classification: low, moderate, high, critical
• Risk score trending: improving, stable, deteriorating
• At-risk driver identification triggers
• New driver monitoring: probationary period scoring
• Recidivist pattern detection
• Predictive risk models: which drivers are likely to have future accidents

Step 2.4 -- Coaching and Remediation

Evaluate:

• Coaching session documentation and tracking
• Remedial training assignment based on behavior patterns
• Coaching effectiveness measurement: behavior change post-coaching
• Progressive discipline integration
• Positive recognition programs
• Coaching frequency targets by risk tier

============================================================ PHASE 3: ACCIDENT ANALYSIS

Step 3.1 -- Accident Recording

Evaluate:

• Accident report data capture: FMCSA-standard fields, first report of injury, photos, statements, police report
• Severity classification: DOT recordable, OSHA recordable, property damage only, near-miss
• Preventability determination process: following ATA guidelines or equivalent
• Root cause analysis methodology: 5-why, fishbone

Step 3.2 -- Accident Trend Analysis

Check for:

• Accident rate calculations: per million miles, per 100 vehicles, per 100 drivers
• Accident type distribution: rear-end, intersection, backing, rollover, pedestrian
• Contributing factor analysis: time of day, day of week, weather, road condition, fatigue
• Geographic hotspot identification
• Seasonal patterns
• Year-over-year trending

Step 3.3 -- Post-Accident Process

Assess:

• Immediate response protocol: drug/alcohol testing triggers, vehicle inspection
• Investigation workflow and timeline
• Corrective action assignment and tracking
• Return-to-duty process
• Modified duty and light-duty management
• Accident review board/committee operations

Step 3.4 -- Cost Impact Analysis

Evaluate:

• Total cost of accidents: vehicle repair, medical, workers' comp, liability, legal, administrative, lost productivity, rental
• Cost attribution: by driver, department, location, accident type
• Insurance impact modeling: premium changes, deductible exposure
• Reserve setting and development tracking for open claims

============================================================ PHASE 4: CSA AND REGULATORY COMPLIANCE

Step 4.1 -- CSA Score Monitoring

Evaluate:

• SMS BASIC score tracking across all seven categories: Unsafe Driving, Hours of Service, Driver Fitness, Controlled Substances/Alcohol, Vehicle Maintenance, Hazardous Materials, Crash Indicator
• Intervention threshold monitoring: percentile rank vs. threshold
• Inspection and violation data feed: FMCSA DataQs integration
• Score projection modeling
• Violation severity weight awareness
• Time-weight decay understanding

Step 4.2 -- Roadside Inspection Management

Check for:

• Inspection result recording and tracking
• Clean inspection rate: no violations found
• Out-of-service rate by category: driver OOS, vehicle OOS
• Inspection location tracking
• DataQ challenge workflow for inaccurate inspection data
• Pre-trip inspection compliance (DVIR)
• Mock inspection programs

Step 4.3 -- Hours of Service Compliance

Assess:

• ELD data integration and monitoring
• HOS violation detection: 11-hour driving, 14-hour window, 30-minute break, 60/70-hour limit
• Unassigned driving time management
• Personal conveyance policy enforcement
• Short-haul exception tracking
• HOS exception utilization: adverse conditions, 16-hour
• Driver log audit workflow

Step 4.4 -- Drug and Alcohol Compliance

Check for:

• Testing program management: pre-employment, random, post-accident, reasonable suspicion, return-to-duty, follow-up
• Random testing pool and selection
• FMCSA Clearinghouse queries: pre-employment and annual
• Substance Abuse Professional (SAP) process tracking
• MRO (Medical Review Officer) result management
• DOT testing rates: minimum 50% random drug, 10% random alcohol for FMCSA

============================================================ PHASE 5: RISK MITIGATION STRATEGIES

Step 5.1 -- Training Programs

Evaluate:

• New driver orientation content and duration
• Defensive driving training: Smith System, LLLC, commentary driving
• Vehicle-specific training: backing, mountain driving, winter driving
• Hazmat training (if applicable)
• Annual refresher requirements
• Training effectiveness measurement
• E-learning and simulation platforms

Step 5.2 -- Safety Technology ROI

Check for:

• Technology impact measurement: accident reduction, severity reduction, near-miss to accident ratio
• Camera system ROI: exoneration savings, coaching impact, deterrence
• Collision avoidance system effectiveness
• Speed limiter impact
• Technology adoption tracking across fleet
• Cost-benefit analysis by technology type

Step 5.3 -- Insurance and Claims Management

Assess:

• Loss run analysis and trending
• Insurance program structure: guaranteed cost, large deductible, self-insured retention
• Claims management workflow
• Subrogation recovery tracking
• Experience modification rate monitoring
• Safety investment impact on premiums

Step 5.4 -- Safety Culture Assessment

Evaluate:

• Safety meeting programs and documentation
• Driver communication platforms
• Anonymous safety concern reporting: near-miss reporting encouragement
• Safety award and recognition programs
• Management safety commitment indicators
• Safety committee structure and effectiveness

============================================================ PHASE 6: DOT AUDIT READINESS

Step 6.1 -- Audit Documentation

Evaluate:

• Driver qualification file completeness: application, MVR, medical certificate, road test, annual review
• Vehicle maintenance file completeness: inspection records, maintenance history, annual inspection
• HOS records retention: 6-month ELD requirement
• Drug and alcohol testing records
• Accident register maintenance

Step 6.2 -- Compliance Gap Analysis

Check for:

• Systematic DQ file audit capability
• Maintenance record audit trail
• HOS compliance rate by driver
• Random testing rate verification
• Vehicle periodic inspection currency
• Hazmat compliance (if applicable): registration, training, shipping papers

============================================================ PHASE 7: WRITE REPORT

Write analysis to docs/fleet-safety-analysis.md (create docs/ if needed).

Include: Executive Summary, Driver Behavior Assessment, Accident Analysis, CSA Performance, HOS Compliance, Drug and Alcohol Program, Risk Mitigation Effectiveness, DOT Audit Readiness, Recommendations with risk reduction estimates.

============================================================ SELF-HEALING VALIDATION (max 2 iterations)

After producing output, validate data quality and completeness:

1. Verify all output sections have substantive content (not just headers).
2. Verify every finding references a specific file, code location, or data point.
3. Verify recommendations are actionable and evidence-based.
4. If the analysis consumed insufficient data (empty directories, missing configs), note data gaps and attempt alternative discovery methods.

IF VALIDATION FAILS:

• Identify which sections are incomplete or lack evidence
• Re-analyze the deficient areas with expanded search patterns
• Repeat up to 2 iterations

IF STILL INCOMPLETE after 2 iterations:

• Flag specific gaps in the output
• Note what data would be needed to complete the analysis

============================================================ OUTPUT

Fleet Safety Analysis Complete

• Report: docs/fleet-safety-analysis.md
• Driver risk tiers: [distribution]
• Accident rate: [per million miles]
• CSA BASIC scores: [summary]
• DOT audit readiness: [score]/10

Summary Table

Area	Status	Priority
Driver Behavior Scoring	[status]	[priority]
Accident Analysis	[status]	[priority]
CSA Compliance	[status]	[priority]
HOS Compliance	[status]	[priority]
Drug & Alcohol	[status]	[priority]
DOT Audit Readiness	[status]	[priority]

NEXT STEPS:

• "Run /fleet-maintenance to assess vehicle condition impact on safety outcomes."
• "Run /fuel-optimization to evaluate how eco-driving aligns with safe driving behaviors."
• "Run /vehicle-routing to ensure routes account for driver fatigue and HOS limits."

DO NOT:

• Do NOT modify any safety records, driver scores, or compliance configurations.
• Do NOT downplay safety violations regardless of their CSA severity weight.
• Do NOT recommend reducing safety technology to cut costs without quantifying risk exposure.
• Do NOT ignore drug and alcohol compliance -- it is the highest-consequence compliance area.
• Do NOT skip DOT audit readiness even if the carrier has not been audited recently.

============================================================ SELF-EVOLUTION TELEMETRY

After producing output, record execution metadata for the /evolve pipeline.

Check if a project memory directory exists:

• Look for the project path in ~/.claude/projects/
• If found, append to skill-telemetry.md in that memory directory

Entry format:

### /fleet-safety — {{YYYY-MM-DD}}
- Outcome: {{SUCCESS | PARTIAL | FAILED}}
- Self-healed: {{yes — what was healed | no}}
- Iterations used: {{N}} / {{N max}}
- Bottleneck: {{phase that struggled or "none"}}
- Suggestion: {{one-line improvement idea for /evolve, or "none"}}

Only log if the memory directory exists. Skip silently if not found. Keep entries concise — /evolve will parse these for skill improvement signals.