A multi-agent AI system that validates requirements against ARP4754A (Guidelines for Development of Civil Aircraft and Systems).
# 2. Create a virtual environment
python -m venv venv
source venv/bin/activate # Linux/macOS
venv\Scripts\activate # Windows
# 3. Install dependencies
pip install -r requirements.txt
# 4. Configure your OpenAI API key
Customize file `.env`
---
## Usage
### Ingest system documents + validate
```bash
python main.py --docs ./system_docs/ --file requirements.txtThe system_docs/ folder can contain any mix of:
- System Description / Concept of Operations (TXT, DOCX, PDF)
- Functional Hazard Assessment (FHA) — failure conditions, DAL levels
- Interface Control Document (ICD) — data formats, protocols, timing
- System Safety Assessment (PSSA/SSA)* — probability targets
- Glossary / Acronym List
- Any other project reference document
python main.py --file requirements.txtpython main.py --docs ./new_docs/ --file reqs.jsonpython main.py --docs ./system_docs/ --clear-store --file reqs.jsonpython main.py --list-docspython main.py --docs ./docs/ --file reqs.json --output report.txt --verbose| Section | Area | Key checks |
|---|---|---|
| §5.2 | Correctness | "shall" usage, compound reqs, HOW vs WHAT, abstraction level |
| §5.3 | Completeness | Unique IDs, acceptance criteria, DAL refs, source/rationale, interface definitions |
| §5.4 | Consistency | Contradictions, terminology, unit conflicts, DAL alignment |
| §5.5 | Verifiability | Quantifiable thresholds, T/A/I/D method mapping, measurability |
| §5.6 | Traceability | Upstream source, FHA linkage, derived req. justification, HW/SW allocation |
# Add to .env
LANGCHAIN_TRACING_V2=true
LANGCHAIN_API_KEY=your-langsmith-key
LANGCHAIN_PROJECT=arp4754-validatorThis gives you a visual trace of every agent call and RAG retrieval in the LangSmith UI.