API Reference
This section provides detailed documentation for all classes and modules in mini-SWE-agent.
Agents
- DefaultAgent - The minimal default agent implementation
- InteractiveAgent - Agent with human-in-the-loop functionality
- TextualAgent - Agent with interactive TUI using Textual
Models
- LitellmModel - Wrapper for LiteLLM models (supports most LLM providers)
- LitellmResponseAPIModel - Specialized model for OpenAI's Responses API
- AnthropicModel - Specialized interface for Anthropic models
- DeterministicModel - Deterministic models for testing
- Model Utilities - Convenience functions for model selection and configuration
Environments
- LocalEnvironment - Execute commands in the local environment
- DockerEnvironment - Execute commands in Docker containers
- SwerexDockerEnvironment - Extended Docker environment with SWE-Rex integration
Run Scripts
Entry points and command-line interfaces:
- Hello World - Simple example usage
- mini - Interactive local execution
- GitHub Issue - GitHub issue solver
- SWE-bench - SWE-bench evaluation script
{% include-markdown "../_footer.md" %}