# Netra Documentation

> Netra is an observability, evaluation, and monitoring platform for AI/LLM applications. This is the official documentation.

## About Netra

Netra provides:
- **Tracing**: Capture LLM calls, agent execution, tool invocations, and multi-step workflows
- **Evaluation**: Systematic quality testing with LLM-as-Judge, rule-based, and code evaluators
- **Monitoring**: Real-time dashboards, alerts, and cost tracking
- **Multi-tenancy**: Native per-tenant tracking for B2B applications

## SDK Installation

Python:
```bash
pip install netra-sdk
```

TypeScript:
```bash
npm install netra-sdk
```

## Quick Start

```python
from netra import Netra

Netra.init(app_name="my-app")
# All OpenAI, Anthropic, LangChain, CrewAI calls are now traced automatically
```

## Documentation Structure

### Getting Started
- /quick-start/Overview - Platform overview and setup
- /quick-start/QuickStart_Tracing - First tracing guide
- /quick-start/QuickStart_Evals - First evaluation guide
- /quick-start/QuickStart_Alerts - Alert setup guide

### Observability
- /Observability/Traces/overview - Tracing concepts
- /Observability/Traces/spans - Span types (Generation, Tool, Agent, Workflow)
- /Observability/Traces/auto-instrumentation - Automatic tracing
- /Observability/Traces/decorators - @agent, @task, @workflow decorators
- /Observability/Traces/manual-tracing - Manual span creation
- /Observability/Agents - Agent-specific observability
- /Observability/Tenants - Multi-tenant tracking
- /Observability/Users - User tracking
- /Observability/Session - Session management

### Evaluation
- /Evaluation/Evaluation-overview - Evaluation concepts
- /Evaluation/Datasets - Test dataset management
- /Evaluation/Evaluators - LLM-as-Judge, rule-based, code evaluators
- /Evaluation/TestRuns - Running evaluation experiments

### Integrations

#### LLM Frameworks
- /Integrations/orchestrators/LangChain
- /Integrations/orchestrators/LangGraph
- /Integrations/orchestrators/CrewAI
- /Integrations/orchestrators/LlamaIndex
- /Integrations/orchestrators/PydanticAI
- /Integrations/orchestrators/LiteLLM
- /Integrations/orchestrators/Haystack
- /Integrations/orchestrators/DSPy
- /Integrations/orchestrators/MCP
- /Integrations/orchestrators/ADK

#### Model Providers
- /Integrations/ai_providers/OPENAI
- /Integrations/ai_providers/ANTHROPIC_CLAUDE
- /Integrations/ai_providers/GEMINI
- /Integrations/ai_providers/AWS_BEDROCK
- /Integrations/ai_providers/MISTRAL
- /Integrations/ai_providers/GROQ
- /Integrations/ai_providers/COHERE
- /Integrations/ai_providers/VERTEXAI
- /Integrations/ai_providers/OLLAMA

#### Vector Databases
- /Integrations/db-docs/Pinecone
- /Integrations/db-docs/CHROMA
- /Integrations/db-docs/Qdrant
- /Integrations/db-docs/Milvus
- /Integrations/db-docs/Weavite

### SDK Reference
- /sdk/overview - SDK overview
- /sdk/python - Python SDK reference
- /sdk/typescript - TypeScript SDK reference

### Analytics
- /analytics-and-dashboards/usage-utilities - Usage APIs (get_tenant_usage, list_traces, list_spans)
- /analytics-and-dashboards/dashboard-query - Dashboard query API

### Cookbooks (End-to-End Tutorials)
- /Cookbooks/pdf-qa-rag-chatbot - RAG pipeline tracing and evaluation
- /Cookbooks/saas-multi-tenant-platform - Multi-tenant B2B observability
- /Cookbooks/langchain-react-agent - LangChain ReAct agent tracing
- /Cookbooks/crewai-content-pipeline - CrewAI multi-agent collaboration

## Key Concepts

### Initialization
```python
from netra import Netra
from netra.instrumentation.instruments import InstrumentSet

Netra.init(
    app_name="my-app",
    instruments={InstrumentSet.OPENAI, InstrumentSet.LANGCHAIN},
    trace_content=True,  # Capture prompts/completions
)
```

### Decorators
```python
from netra.decorators import agent, task, workflow

@workflow(name="my-workflow")
def my_workflow():
    pass

@agent(name="my-agent")
def my_agent():
    pass

@task(name="my-task")
def my_task():
    pass
```

### Manual Spans
```python
from netra import Netra, SpanType

with Netra.start_span("my-operation", as_type=SpanType.TOOL) as span:
    span.set_attribute("key", "value")
    # Your code here
```

### Multi-Tenant Tracking
```python
Netra.set_tenant_id("customer-123")
Netra.set_user_id("user-456")
Netra.set_session_id("session-789")
```

### Usage APIs
```python
usage = Netra.usage.get_tenant_usage(
    tenant_id="customer-123",
    start_time="2026-01-01T00:00:00.000Z",
    end_time="2026-01-31T23:59:59.000Z",
)
```

## Links

- Website: https://getnetra.ai
- Dashboard: https://app.eu.getnetra.ai/login
- Python SDK: https://pypi.org/project/netra-sdk/
- TypeScript SDK: https://www.npmjs.com/package/netra-sdk
- GitHub: https://github.com/KeyValueSoftwareSystems/netra-sdk-py