Quality engineering skills
for AI agents

Quality Engineering Skills is an open-source library of structured quality methodology files for AI agents. It gives any AI — ChatGPT, Claude, Gemini — the same domain knowledge and validation rules that a senior quality engineer would apply.

Instead of asking an AI "how do I run a 5-Why?" and getting generic advice, a QES-powered agent asks you for evidence at each step, challenges circular reasoning, and tells you whether your root cause chain is strong enough to close the 8D.

No — that's the point. The agents guide you through each methodology step by step and explain what a good answer looks like when they challenge yours. You learn the methodology by doing it under guidance.

That said, practitioners with quality engineering background will get more value faster, because they can recognise when the agent's guidance matches what they know from experience.

Asking ChatGPT "how do I do 8D?" gives you an explanation. QES gives an AI the methodology to run a real 8D with you — one discipline at a time, with gate validation.

The key differences:

• D3 containment is rejected if it's future-tense — "we will inspect" gets challenged immediately
• D4 requires two root causes — occurrence AND escape — not just one
• "Human error" as a root cause is always challenged — the agent asks why human error was possible
• The agent knows the OEM-specific 8D formats (Ford, BMW, VW/Audi, Stellantis) and their SQE requirements

Yes. Every skill is grounded in specific standard editions:

• ISO 9001:2015 — problem solving, NCR, CAR, audits
• IATF 16949:2016 — supplemental audit, PPAP, special characteristics
• AIAG-VDA FMEA Handbook 2019 — Action Priority, 7-step PFMEA/DFMEA method
• AIAG PPAP 4th Edition (2006) — 18 elements, 5 submission levels, OEM overrides
• AIAG MSA 4th Edition (2010) — Gauge R&R, %GRR, ndc, attribute MSA
• AIAG SPC 2nd Edition (2005) — chart selection, Western Electric rules, capability indices
• VDA 6.3 4th Edition (2023) — P1–P7 process audit, weighted scoring

All methodology claims cite the specific standard clause or section. Generic content is rejected in pull requests.

The 8D Report Writing skill and the 8D Coach agent cover:

• Ford — 8D form via Covisint / GSDB portal, D4 root cause requirements
• BMW — Global 8D (G8D) via SupplierPortal (SIM)
• VW / Audi — QMSS (Lieferantenselbstauskunft) via SupplyOn
• Stellantis — 8D via CQMS, SQE approval for H-AP acceptance
• Mercedes-Benz — 8D / ATC via SupplyOn

Each OEM has different D4 root cause depth requirements and D7 documentation standards. The skill knows these differences.

The 2019 AIAG-VDA joint handbook replaced the old Risk Priority Number (RPN = S×O×D) with Action Priority (AP) — H/M/L. Key changes:

• Severity = 9 or 10 is always High Priority, regardless of occurrence or detection — this was not true with RPN
• AP=H items require mandatory action or a formally documented engineering justification — "we accept this risk" is not sufficient without documented sign-off
• The 7-step method (Structure → Function → Failure → Risk → Optimization → Results → Documentation) replaced the open-ended approach

The pfmea-process and action-priority-ap skills implement the 2019 method, not the old handbook.

Yes. The core methodologies — 8D, 5-Why, PFMEA, NCR writing, ISO 9001 audit — apply across automotive, electronics, aerospace, medical devices, and general manufacturing. The OEM-specific sections are clearly marked.

The metadata on each skill includes an industries field. All 22 base skills include automotive, electronics, aerospace, medical, general. Only the OEM-specific reference files are automotive-only.

The iatf-16949-audit skill covers all 16 supplemental requirements, including the three most commonly non-conforming:

• §4.3.2 Customer-specific requirements (CSR) — register, review evidence, implementation traceability
• §6.1.2.3 Contingency plans — documented, reviewed, tested — not just written
• §8.3.3.3 Special characteristics — consistent SC symbols across all documents (drawing, DFMEA, PFMEA, CP, WI)

The three mandatory IATF audit streams (QMS audit, manufacturing process audit, product audit) are also covered.

Yes. Quality Engineering Skills is released under the MIT License. Use it, embed it, build SaaS products on top of it — the only requirement is keeping the copyright notice in any distribution.

The 8D Evaluator is permanently free — it will never move to a paid tier.

All skills must be grounded in actual standard methodology — generic AI text is rejected. To contribute:

1. Contact @RBraga01 for collaborator access
2. Create a branch: feat/your-skill-name
3. Follow the SKILL.md format (frontmatter + required sections + Output Format + reference files)
4. Include all document control fields: status, created, reviewed_by, standard_edition
5. Open a PR with: skill name, standard it implements, your domain expertise

Skills from practitioners are prioritised. The Skill Auditor agent scores every submission — minimum score 9/10 to merge.