cli-reference.md

CLI Reference

Installation

pip install trusteval

Commands

trusteval evaluate

Run an LLM evaluation.

trusteval evaluate \
  --provider openai \
  --model gpt-4 \
  --industry healthcare \
  --pillars bias,hallucination,pii,toxicity \
  --output report.pdf \
  --verbose

Options:

• --provider — LLM provider (openai, anthropic, gemini, huggingface)
• --model — Model name
• --industry — Industry module (healthcare, bfsi, retail, legal)
• --pillars — Comma-separated pillar list (default: all)
• --output — Output report path
• --prompt — Single prompt for quick evaluation
• --verbose — Enable verbose output

trusteval compare

Compare two models side-by-side.

trusteval compare \
  --provider-a openai --model-a gpt-4 \
  --provider-b anthropic --model-b claude-3-opus \
  --industry healthcare

trusteval providers

trusteval providers list          # List all supported providers
trusteval providers test --provider openai    # Test connectivity
trusteval providers configure --provider openai  # Configure API key

trusteval report

trusteval report generate \
  --input results.json \
  --format pdf \
  --output audit-report.pdf

trusteval dashboard

trusteval dashboard start            # Launch with browser
trusteval dashboard start --port 8080 --no-browser

trusteval config

trusteval config set industry healthcare
trusteval config show
trusteval config reset