Arena Reference
Complete technical specifications and reference materials for PromptArena.
Quick Links
Section titled “Quick Links”Complete command-line interface reference with all flags and options.
YAML configuration file structure and all available options.
All available assertion types for validating LLM responses.
Built-in validators for checking response quality and compliance.
Test scenario file structure and specification.
Report generation formats (HTML, JSON, JUnit, Markdown).
Complete duplex streaming configuration for voice testing scenarios.
Reference vs. How-To
Section titled “Reference vs. How-To”This is reference documentation - dry, factual, technical specifications.
Looking for task-oriented guides? See:
- Arena How-To Guides - Accomplish specific tasks
- Arena Tutorials - Learn by building
Quick Reference Tables
Section titled “Quick Reference Tables”Command Summary
Section titled “Command Summary”| Command | Purpose |
|---|---|
promptarena run | Execute test scenarios |
promptarena config-inspect | Validate configuration |
promptarena debug | Debug configuration loading |
promptarena prompt-debug | Test prompt rendering |
promptarena render | Generate reports from results |
Common Assertions
Section titled “Common Assertions”| Assertion | Purpose |
|---|---|
not_empty | Response is not empty |
contains | Response contains text |
matches | Response matches regex |
tool_called | Specific tool was invoked |
max_tokens | Token count within limit |
semantic_similarity | Meaning matches expected |
Output Formats
Section titled “Output Formats”| Format | Use Case |
|---|---|
| JSON | Machine processing, APIs |
| HTML | Human-readable reports |
| JUnit | CI/CD integration |
| Markdown | Documentation, sharing |
API Stability
Section titled “API Stability”Arena reference documentation follows semantic versioning:
- Stable: CLI commands, configuration schema
- Beta: Advanced assertions, custom validators
- Experimental: New features marked explicitly
Getting Help
Section titled “Getting Help”- How-To Guides: Task-oriented documentation
- Tutorials: Learning-oriented guides
- Explanations: Conceptual documentation
- Issues: GitHub Issues