Installation
# Global install (optional) npm install -g llmverify # Or use npx (recommended) npx llmverify <command>
Quick Reference
| Command | Description |
|---|---|
| llmverify run | ★ Master command — run all engines with presets |
| llmverify wizard | ★ Interactive setup wizard |
| llmverify verify | Run multi-engine verification (default) |
| llmverify presets | List available preset configurations |
| llmverify benchmark | Benchmark latency across presets |
| llmverify engines | List all verification engines |
| llmverify explain <engine> | Explain how an engine works |
| llmverify adapters | List available LLM adapters |
| llmverify doctor | Check system health |
| llmverify init | Initialize config file |
| llmverify privacy | Show privacy guarantees |
| llmverify info | Show package information |
| llmverify tutorial | Show usage examples |
llmverify run ★
Master command that runs all verification engines with preset configurations. Recommended for most use cases.
llmverify run [content] [options]
Options
| Option | Description | Default |
|---|---|---|
| -f, --file <path> | Read content from file | - |
| -p, --preset <mode> | Preset: dev, prod, strict, fast, ci | dev |
| --prompt <text> | Original prompt for classification | - |
| --input <text> | User input to check for injection | - |
| -o, --output <format> | Output: text, json, summary | text |
| --parallel | Run engines in parallel | true |
| --no-parallel | Run engines sequentially | - |
Presets
| Preset | Description | Speed | Thoroughness |
|---|---|---|---|
| dev | Development — balanced output | ●●●○○ | ●●●●○ |
| prod | Production — optimized for speed | ●●●●● | ●●●○○ |
| strict | Strict — maximum scrutiny | ●●○○○ | ●●●●● |
| fast | Fast — minimal checks | ●●●●● | ●●○○○ |
| ci | CI — optimized for pipelines | ●●●●○ | ●●●●○ |
Examples
# Development mode (recommended for starting) npx llmverify run "Your AI output" --preset dev # Production mode npx llmverify run "Your AI output" --preset prod # From file with strict checking npx llmverify run --file output.txt --preset strict # JSON output for CI/CD npx llmverify run "Your AI output" --preset ci --output json # With original prompt for classification npx llmverify run "The capital is Paris" --prompt "What is the capital of France?"
Exit Codes (CI/CD)
| Code | Meaning | Action |
|---|---|---|
| 0 | Low risk | Allow |
| 1 | Moderate risk | Review |
| 2 | High/Critical risk | Block |
llmverify wizard ★
Interactive setup wizard that guides you through configuration.
npx llmverify wizard # Creates .llmverifyrc.json with your preferences # Asks about: engines, thresholds, output format, presets
llmverify doctor
Check system health and verify installation.
npx llmverify doctor # Checks: # ✓ Node.js version # ✓ Package installation # ✓ Configuration file # ✓ Engine availability # ✓ Disk space for audit logs
llmverify benchmark
Benchmark latency across all presets.
npx llmverify benchmark # Output: # Preset | Avg Latency | Engines # dev | 25ms | 5/5 # prod | 12ms | 3/5 # strict | 45ms | 5/5 # fast | 5ms | 2/5 # ci | 18ms | 4/5