instructor - Extract validated structured data from LLMs
Extracts structured data from LLM responses and validates it with Pydantic, including retries and streaming support.
Tags
Updated: 2026-10-02Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Extract structured data
- Validate outputs with Pydantic
- Retry failed extractions
- Parse typed JSON
- Stream partial results
- Support OpenAI and Anthropic
Inputs
- Source text or prompt
- Pydantic response model
- LLM provider credentials
- Model and request configuration
Outputs
- Validated structured objects
- Validation errors
- Partial streamed objects
- Streamed iterable items
Requirements
- Python environment
- Installed instructor, Pydantic, OpenAI, and Anthropic packages
- Provider API key or local Ollama server
- Linux, macOS, or Windows