Skip to content

Latest commit

 

History

History
37 lines (25 loc) · 2.17 KB

File metadata and controls

37 lines (25 loc) · 2.17 KB

Internal Examples

This directory contains comprehensive examples that test all features of the Braintrust Go SDK. These examples are primarily used for:

  • SDK development and validation - Ensuring all features work correctly
  • Regression testing - Catching breaking changes during development
  • Feature coverage - Demonstrating every SDK capability

Purpose

Unlike the customer-facing examples in the parent directory, these examples are not optimized for learning. They are "kitchen sink" examples that exercise many features at once to verify SDK functionality.

Examples

Provider Kitchen Sinks

Comprehensive examples testing all features for each AI provider:

  • anthropic/ - All Anthropic features (messages, tools, streaming, citations, thinking, vision)
  • openai-v1/ - All OpenAI v1 features (chat, streaming, tools, vision)
  • openai-v2/ - All OpenAI v2 features (Responses API, reasoning, conversations)
  • genai/ - All Gemini features (text, system instruction, multi-turn, thinking, streaming, functions, safety, JSON mode, multimodal)
  • genkit/ - Genkit generation with manual middleware, config, and streaming coverage
  • langchaingo/ - All LangChainGo features (simple, multi-turn, chains, tools, agents, retriever, streaming, system prompt, temperature, max tokens, prefill, stop sequences, long context, metadata)

Specialized Testing

  • langchaingo-anthropic/ - LangChainGo with Anthropic provider (uses forked langchaingo)
  • functions/ - Functions API usage (loading tasks/scorers with FunctionOpts)
  • rewrite/ - Manual tracing and evaluator API testing
  • email-evals/ - Realistic eval example with complex scoring
  • eval-updates/ - Testing Update option for appending to experiments
  • temporal/ - Temporal workflow distributed tracing (worker + client)

For Learning

If you're learning the Braintrust SDK, start with the examples in the parent directory (../). Those examples are concise, focused, and designed for customer education.