Welcome to the traceAI documentation. This guide covers installation, configuration, and usage of traceAI for instrumenting your AI applications.
- Installation - How to install traceAI
- Python Quickstart - Get started with Python
- TypeScript Quickstart - Get started with TypeScript
- TraceConfig - Privacy and data redaction settings
- Environment Variables - All configuration options
- Evaluation Tags - AI evaluation system
- Core Concepts - Understanding fi_instrumentation
- Context Managers - Adding metadata to spans
- fi-core Package - Core TypeScript library
- Instrumentations - Framework instrumentations
- Python Examples - Python code walkthroughs
- TypeScript Examples - TypeScript code walkthroughs
| Resource | Description |
|---|---|
| Main README | Project overview and supported frameworks |
| Python SDK | Python-specific documentation |
| External Docs | Full documentation site |
| Cookbooks | Step-by-step tutorials |
traceAI provides drop-in OpenTelemetry instrumentation for AI applications. It automatically captures:
- LLM Calls: Prompts, completions, token usage, model parameters
- Agent Workflows: Task execution, tool calls, reasoning steps
- Retrieval: Document fetching, embeddings, vector searches
- Errors: Detailed error context and stack traces
Your Application
│
▼
┌─────────────────┐
│ traceAI SDK │ ← Instruments your framework
└────────┬────────┘
│
▼
┌─────────────────┐
│ OpenTelemetry │ ← Standard tracing API
└────────┬────────┘
│
▼
┌─────────────────┐
│ OTLP Exporter │ ← Sends traces
└────────┬────────┘
│
▼
┌─────────────────┐
│ Future AGI / │ ← Visualize & analyze
│ Any OTel │
│ Backend │
└─────────────────┘
traceAI captures different types of spans:
| Span Kind | Description | Example |
|---|---|---|
LLM |
Language model calls | Chat completions, text generation |
AGENT |
Agent orchestration | CrewAI tasks, AutoGen agents |
TOOL |
Tool/function calls | API calls, code execution |
RETRIEVER |
Document retrieval | RAG queries, vector search |
EMBEDDING |
Embedding generation | Text to vector conversion |
CHAIN |
Workflow chains | LangChain chains, pipelines |
RERANKER |
Result reranking | Search result ordering |
- GitHub Issues - Report bugs or request features
- Slack Community - Ask questions and get help
- Stack Overflow - Community Q&A