██╗    ██╗███████╗ █████╗ ██╗   ██╗███████╗███████╗ █████╗ ██╗   ██╗██╗  ████████╗
██║    ██║██╔════╝██╔══██╗██║   ██║██╔════╝██╔════╝██╔══██╗██║   ██║██║  ╚══██╔══╝
██║ █╗ ██║█████╗  ███████║██║   ██║█████╗  █████╗  ███████║██║   ██║██║     ██║
██║███╗██║██╔══╝  ██╔══██║╚██╗ ██╔╝██╔══╝  ██╔══╝  ██╔══██║██║   ██║██║     ██║
╚███╔███╔╝███████╗██║  ██║ ╚████╔╝ ███████╗██║     ██║  ██║╚██████╔╝███████╗██║
 ╚══╝╚══╝ ╚══════╝╚═╝  ╚═╝  ╚═══╝  ╚══════╝╚═╝     ╚═╝  ╚═╝ ╚═════╝ ╚══════╝╚═╝

"We weave through your architecture to find where it will fault."

---

What is WeaveFault?

WeaveFault is an automated FMEA (Failure Mode & Effects Analysis) generator that:

1. Reads your architecture diagrams — PNG, SVG, drawio, PDF
2. Builds a dependency graph — using NetworkX, with centrality metrics
3. Simulates failure propagation — BFS cascade through the directed graph
4. Generates structured FMEA documents — AIAG, IEC 60812, MIL-STD-1629, ISO 26262
5. Shows full AI reasoning chains — every RPN score is fully auditable

WeaveFault is diagram-native, git-integrated, cross-domain, and glass-box.

---

The 5 Pillars

1. Topology-First Input

Your architecture diagram is your failure map. WeaveFault extracts every component, edge, and data flow — then reasons about failure from the structure of the system itself, not from a blank spreadsheet.

2. Cascade Failure Graph

Failures don't happen in isolation. WeaveFault simulates propagation through your dependency graph using directed BFS, computing blast radius, critical paths, and single points of failure (SPOFs) before a single FMEA row is written.

3. Living FMEA (Git-Native, Diffs on Commit)

FMEA documents are checked into git alongside diagrams. When diagrams change, WeaveFault regenerates and produces a structured diff report — showing exactly which failure modes are new, removed, or changed. Your FMEA evolves with your system.

4. Cross-Domain Reasoning

WeaveFault supports cloud infrastructure, embedded firmware, mechanical systems, and hybrid architectures. Domain-specific failure modes, standards, and severity contexts are configurable via config/domains.yaml.

5. Glass-Box AI

Every FMEA row includes the full LLM reasoning chain that produced it. Severity, Occurrence, and Detection scores are not black boxes — they are justified, auditable, and can be challenged in code review.

---

Installation

pip install weavefault

Or from source:

git clone https://github.com/Hexcore2x/weavefault.git
cd weavefault
pip install -e .

Set up your environment:

cp .env.example .env
# Edit .env and add your API keys

---

Quick Start

# Generate FMEA from a diagram
weavefault generate \
  --diagram ./diagrams/auth_service.png \
  --domain  cloud \
  --output  ./fmea/ \
  --format  excel

# Diff two FMEA snapshots
weavefault diff \
  --before ./fmea/v1.md \
  --after  ./fmea/v2.md \
  --output ./fmea/diff_report.md

# Watch diagram folder and auto-regenerate on change
weavefault watch \
  --diagram ./diagrams/ \
  --output  ./fmea/ \
  --branch  main

---

CLI Reference

weavefault generate

Parse a diagram, run cascade simulation, generate FMEA, and export.

Option	Type	Default	Description
--diagram	PATH	required	Input diagram file or directory
--domain	TEXT	cloud	System domain (cloud/embedded/mechanical/hybrid)
--standard	TEXT	IEC_60812	FMEA standard to apply
--output	PATH	required	Output directory
--format	CHOICE	both	Export format: excel, markdown, both
--provider	TEXT	env	LLM provider (anthropic/openai)
--model	TEXT	env	LLM model name
--verbose	FLAG	false	Enable verbose logging

weavefault diff

Compare two FMEA snapshots and produce a diff report.

Option	Type	Default	Description
--before	PATH	required	First FMEA snapshot (.weavefault.json or .md)
--after	PATH	required	Second FMEA snapshot
--output	PATH	required	Output path for diff report (.md)

weavefault watch

Monitor a diagram directory and regenerate FMEA on every change.

Option	Type	Default	Description
--diagram	PATH	required	Directory to watch
--output	PATH	required	Output directory
--interval	INT	2	Poll interval in seconds

---

How It Works

Architecture Diagram
        │
        ▼
┌─────────────────┐
│  DiagramParser  │  Vision LLM extracts components + edges
│  (PNG/SVG/etc)  │  → DiagramGraph (Pydantic model)
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  GraphBuilder   │  NetworkX DiGraph
│  + Centrality   │  Betweenness, in/out-degree, SPOFs
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│ CascadeSimulator│  BFS propagation per node
│  blast radius   │  → CascadeChain per component
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│ FMEAGenerator   │  Graph-aware LLM prompts
│  + RAG context  │  3-5 failure modes per component
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  RPNScorer      │  Severity × Occurrence × Detection
│  + ReasonChain  │  Full audit trail per row
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  Exporters      │  .xlsx / .md / .weavefault.json
│  + DiffEngine   │  Git-native diff reports
└─────────────────┘

---

Supported Domains

Domain	Description	Example Systems
cloud	Cloud-native microservices, APIs, databases	AWS, GCP, K8s architectures
embedded	Real-time firmware, sensors, actuators	IoT, automotive ECU, RTOS
mechanical	Physical systems, manufacturing, materials	CNC, pneumatics, structural
hybrid	Multi-domain systems	Connected vehicles, smart factories

---

Supported Standards

Standard	Full Name	Domain	RPN High-Risk
AIAG_FMEA4	AIAG FMEA-4	Automotive	≥ 100
IEC_60812	IEC 60812:2018	General / Industrial	≥ 200
MIL_STD_1629	MIL-STD-1629A	Defense / Aerospace	≥ 200
ISO_26262	ISO 26262:2018	Functional Safety / Automotive	ASIL-based

---

Uploading to GitHub Later

When you are ready to push this project to GitHub:

git remote add origin https://github.com/Hexcore2x/weavefault.git
git add .
git commit -m "feat: initial WeaveFault project"
git push -u origin main

The GitHub Actions workflow at .github/workflows/weavefault-ci.yml will activate automatically on push and begin auto-generating FMEA on diagram changes.

---

Contributing

1. Fork the repository
2. Create a feature branch: git checkout -b feat/my-feature
3. Make your changes with tests
4. Run black src/ tests/ and ruff check src/ tests/
5. Submit a pull request

---

License

MIT License — see LICENSE for details.

---

WeaveFault is built on the belief that every architecture diagram is a failure map waiting to be read.