> ## Documentation Index
> Fetch the complete documentation index at: https://docs.trulayer.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# TruLayer Documentation

> The truth layer for AI systems — observability, evaluation, debugging, feedback, and control for LLM applications.

TruLayer is a reliability platform for AI applications. Instrument your app with one of our SDKs, and you get end-to-end traces of every prompt, tool call, retrieval, and response — plus automated evaluations, user feedback capture, and real-time failure detection.

## Start here

<CardGroup cols={2}>
  <Card title="Quickstart" icon="rocket" href="/quickstart">
    Install the SDK and send your first trace in under 10 minutes.
  </Card>

  <Card title="Core concepts" icon="diagram-project" href="/concepts/overview">
    Understand traces, spans, sessions, evals, feedback, and metrics.
  </Card>
</CardGroup>

## SDKs

<CardGroup cols={3}>
  <Card title="Python SDK" icon="python" href="/sdks/python/overview">
    `pip install trulayer` — auto-instrumentation for OpenAI, Anthropic, LangChain.
  </Card>

  <Card title="TypeScript SDK" icon="js" href="/sdks/typescript/overview">
    `npm install @trulayer/sdk` — zero-dependency, Node/Edge/Bun compatible.
  </Card>

  <Card title="Go SDK" icon="golang" href="/sdks/go/overview">
    `go get github.com/trulayer/client-go` — zero-dependency, goroutine-safe.
  </Card>
</CardGroup>

## Reference

<CardGroup cols={2}>
  <Card title="HTTP API" icon="terminal" href="/api-reference/introduction">
    Every endpoint, request shape, and response — auto-generated from our OpenAPI spec.
  </Card>

  <Card title="Dashboard" icon="chart-line" href="https://app.trulayer.ai">
    Traces, evals, metrics, failures, and team settings.
  </Card>
</CardGroup>

## What TruLayer gives you

* **Observability** — full trace capture (prompt → tools → outputs), session replay, failure classification.
* **Evaluations** — automated eval pipelines with metrics for correctness, hallucination, and safety; regression testing against datasets.
* **Debugging** — step-by-step reasoning inspection, tool-call validation, error clustering.
* **Feedback** — user feedback ingestion, auto-labeling, continuous dataset generation.
* **Control** — real-time failure detection, auto-retry and fallback strategies, dynamic model routing.

<Note>
  **New here?** The [Quickstart](/quickstart) gets you from zero to a trace in the dashboard in under 10 minutes. If you prefer to understand the model before writing code, start with [Core concepts](/concepts/overview).
</Note>