hil-test-tool 0.1.0

Generate structured HIL/SIL/MIL test cases from requirements documents using local AI — NDA safe, 100% offline

hil-test-tool

Generate structured HIL/SIL/MIL test cases from requirements documents using local AI — 100% offline, NDA safe.

pip install hil-test-tool

---

Why

Writing HIL test cases manually from requirements documents is tedious and error-prone. This tool reads your requirements file, sends each requirement to a local AI model (via Ollama), and generates a structured Excel test matrix in the standard HIL test case format.

100% local inference — your requirements never leave your machine.

---

Requirements

• Python 3.9+
• Ollama running locally
• Mistral pulled: ollama pull mistral

---

Install

pip install hil-test-tool

---

Usage

# Generate test matrix from a requirements document
hil-tool generate GPS_Requirements.docx

# Use a different model
hil-tool generate requirements.docx --model mixtral

# Generate for specific requirements only
hil-tool generate requirements.docx --req HIL-RQ1 --req HIL-RQ5

# Custom output filename
hil-tool generate requirements.docx --output GPS_TestMatrix.xlsx

# Custom sheet title
hil-tool generate requirements.docx --title "GPS Block — HIL Test Cases"

# List available Ollama models
hil-tool models

# Show version and config
hil-tool info

---

Output format

Single sheet: TestCases_SystemRequirement

Test Case ID	Requirement Ref	Description	Variant Conditions (Inputs)	Expected Output	Pass Criteria	Result
TC-GPS-001	HIL-RQ1	Verify initialization...	Pre: ... Inputs: ... Monitor: ...	Expected value...	Within ±0.00001°

Test cases are grouped by requirement with GROUP header rows. Result column is left blank for manual fill after test execution.

---

Supported input formats

• .docx — Word requirements documents
• .xlsx — Excel requirements sheets

---

Model recommendations

Model	Quality	Speed	Use case
mistral (default)	Good	Fast	Standard use
mixtral	Excellent	Slow	Complex requirements
llama3	Fair	Fast	Fallback

---

Design principles

• Never guess — the AI only uses signal names and values explicitly stated in the requirements
• Skip, don't crash — if a requirement fails after 3 attempts, it is skipped and reported
• NDA safe — Ollama runs locally, nothing is sent to any cloud API

---

Planned (v0.2)

• Test vector output for manual testing
• Python pytest script generation
• dSPACE integration
• ReqIF format support