API Reference

Complete API reference for SynapseKit classes and methods.

📄️ LLM API Reference

BaseLLM

📄️ RAGPipeline API Reference

RAGConfig

📄️ StateGraph API Reference

StateGraph

📄️ Agents API Reference

AgentConfig

📄️ Memory API Reference

SynapseKit provides 9 memory backends for conversation history management. All implement BaseMemory.

📄️ Retriever API Reference

SynapseKit provides 16 retriever strategies. All implement BaseRetriever.

📄️ Vector Store API Reference

VectorStore (ABC)

📄️ Evaluation API Reference

SynapseKit's evaluation framework measures RAG and agent quality using LLM-as-judge metrics.

📄️ Observability API Reference

SynapseKit provides token tracing, cost tracking, budget enforcement, OpenTelemetry export, and distributed tracing.

📄️ Structured Output API Reference

Provider-agnostic Pydantic validation for LLM responses. StructuredOutput wraps any LLM, validates its JSON response against a Pydantic v2 model, and automatically retries with a corrective prompt on validation failure. Streaming support yields incremental JSON chunks alongside retry and result events.

📄️ Continuous Trainer API Reference

Closed-loop fine-tuning pipeline that collects production feedback, converts it into training data, runs a fine-tuning job, and gradually rolls out the resulting model through A/B testing.

📄️ SelfHealingRAG API Reference

SelfHealingRAG retries RAG retrieval with alternate strategies when answer faithfulness falls below a quality threshold.

📄️ CostQualityRouter API Reference

CostQualityRouter is a learning-based LLM router that explores candidate models, then exploits the cheapest one meeting a learned quality threshold.

📄️ Voice Pipeline API Reference

Real-time voice agents in Python: VAD, streaming STT, LLM, and streaming TTS composed into a single VoicePipeline with interruption handling.