Version 1.0 — A comprehensive AI-powered analysis skill for automotive safety-critical systems engineering. Works with Claude Code, ChatGPT (Custom GPTs), Cursor, Gemini, or any AI tool that accepts system prompts.

Covers the full automotive V-model lifecycle — from requirements quality analysis through code compliance, ADAS classification, functional safety, and architecture review.

• ISO 26262 — Functional safety for automotive E/E systems (ASIL A-D)
• ISO 21448 (SOTIF) — Safety of the Intended Functionality
• ISO 21434 — Automotive cybersecurity engineering
• SAE J3016 — Driving automation levels (L0-L5)
• MISRA-C:2012 — C coding standard for safety-critical embedded systems
• AUTOSAR — Classic Platform + Adaptive Platform architectures
• INCOSE GfWR — 42 requirements quality rules, 14 quality characteristics
• EARS — 6 requirement syntax patterns (Ubiquitous, Event, State, Unwanted, Optional, Complex)
• UNECE R157 — Automated Lane Keeping System regulation

# Clone into your skills directory
git clone https://github.com/duonghvu/automotive-syseng.git ~/.claude/skills/automotive-syseng

# Use it
claude
> /automotive-syseng ears

1. Create a new Custom GPT in ChatGPT
2. Copy the contents of SKILL.md into the Instructions field
3. Upload the files in references/ as Knowledge files
4. Save and use

Copy the contents of SKILL.md and append the relevant references/*.md files into your system prompt or rules file (.cursorrules, etc.).

Just paste SKILL.md into any chat as your system prompt, then attach the relevant reference file from references/ for the specific mode you need.

automotive-syseng/
├── SKILL.md                          # Main skill definition (~7.5KB, ≤10KB per Claude standard)
├── README.md                         # This file
├── CONTRIBUTING.md                   # How to contribute
├── LICENSE                           # MIT
└── references/                       # Deep knowledge, loaded on demand
    ├── ears-patterns.md              # All 6 EARS templates with automotive examples
    ├── incose-quality-rules.md       # INCOSE GfWR — all 42 rules (R1-R42), 14 characteristics
    ├── requirements-quality-indicators.md  # 16 problem types, indicator word lists, 1-5 scoring
    ├── misra-c-rules.md              # MISRA-C:2012 mandatory/required/advisory rules
    ├── v-w-model.md                  # V-model phases + W-model parallel test engineering
    ├── adas-levels.md                # SAE J3016 L0-L5 with sensor/requirements tables
    ├── iso-26262-asil.md             # ASIL matrix, decomposition, safety case structure
    └── autosar-sotif.md              # Classic vs Adaptive + SOTIF triggering conditions

Input:

The system shall be able to quickly detect obstacles and respond appropriately.

Output:

Score: 1/5 (Red)
EARS Pattern: Non-conforming
Issues:
  [P1] Multiple problems triggered
  [P4] Vague Word: "appropriately" (R7)
  [P7] Superfluous Infinitive: "be able to" (R10)
  [P10] Non-Specific Temporal: "quickly" (R35)

Suggested Rewrite:
  "When an obstacle is detected within 50 m of the vehicle, the ADAS controller
   shall apply braking force to achieve a deceleration of at least 5 m/s² within
   100 ms."

uint8_t calculate(uint16_t a, uint16_t b) {
    return a + b;  // Violation: Rule 10.3 implicit narrowing
}

Output:

[Rule 10.3] Required - Line 2
Issue: Implicit narrowing conversion from uint16_t to uint8_t
Fix: return (uint8_t)((a + b) & 0xFFu);  // explicit cast + mask

Input: "Our system controls steering and braking on highways with driver supervision required."

Output: SAE Level 2 — Partial Driving Automation. Checklist verified for: dual-axis control, DMS requirement, sensor fusion, ODD boundary detection.

This skill encodes publicly available standards and best practices from:

• INCOSE Guide for Writing Requirements (4th edition)
• MISRA-C:2012 + Amendments (publicly documented rules)
• SAE J3016 (2021 revision) — Levels of Driving Automation
• ISO 26262:2018 (2nd edition) — Road vehicles functional safety
• ISO 21448:2022 — Safety of the Intended Functionality
• AUTOSAR Classic + Adaptive Platform specifications
• EARS templates by Alistair Mavin (Rolls-Royce, originally)

No proprietary tool data, no copyrighted standard text — only the publicly known rule structures, indicator patterns, and methodologies.

Initial public release. Includes 8 analysis modes, 8 reference files, and follows the Claude Code skill authoring standard (SKILL.md ≤10KB, references loaded on demand, practitioner voice, multi-mode workflows).

Planned for future versions:

• v1.1 — Python helper scripts for batch requirement analysis (CSV/ReqIF input)
• v1.2 — ASPICE process compliance mode
• v1.3 — ISO 21434 cybersecurity TARA mode
• v1.4 — UNECE R155/R156 cybersecurity + software update regulation checks
• v1.5 — Hardware safety analysis (FMEDA, FTA quantitative)
• v2.0 — Multi-language MISRA-C++ support, AUTOSAR ARXML parser

Contributions welcome! See CONTRIBUTING.md for:

• Reporting issues or false positives
• Requesting new features (open an Issue with the enhancement label)
• Submitting Pull Requests for new modes, reference files, or examples
• Adding examples for specific automotive subsystems
• Translating to other languages

MIT — see LICENSE. Free for commercial and non-commercial use.

This skill provides AI-assisted analysis as a productivity aid. It is not a certified tool and does not replace:

• Certified static analyzers (Polyspace, PC-lint, QA-C, LDRA) for MISRA-C compliance evidence
• Commercial requirements management tools (DOORS, Polarion, Jama) for traceability of record
• Independent safety assessor review for ISO 26262 certification

Use this skill for early-stage analysis, training, prototyping, and as a complement to certified processes — not as a substitute for them.

• INCOSE for the Guide for Writing Requirements
• Alistair Mavin et al. for the EARS notation
• The Anthropic team for the Claude Code skill format