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

# The 5 Layers Overview

> The L · I · A · T · E Sovereign Architecture

## The LIATE Specification

OpenLiate decomposes agent design into five isolated, modular layers. In `agent.json`, each layer corresponds to a top-level key:

```json agent.json theme={"dark"}
{
  "L": "sarvam/sarvam-105b",
  "I": { "memory": "sessions/agent-data" },
  "A": { "name": "analyst", "maxSteps": 10 },
  "T": [ "mcp:fetch", "verify_identity" ],
  "E": { "SARVAM_API_KEY": "$SARVAM_API_KEY" }
}
```

***

> [card] [\[L\] Large Models Layer](https://docs.openliate.tech/layers/large-models)
>
> Provider abstraction supporting Indian sovereign models (Sarvam AI), Ollama, Anthropic, and OpenAI.

> [card] [\[I\] Integration & Memory Layer](https://docs.openliate.tech/layers/integration-memory)
>
> Multi-tier session persistence: In-Memory, File-backed JSON, and Vector DB integration.

> [card] [\[A\] Agent Harness & Loop Layer](https://docs.openliate.tech/layers/agent-loop)
>
> Autonomous ReAct execution loop (Thought $\rightarrow$ Action $\rightarrow$ Observation $\rightarrow$ Answer) with loop guardrails.

> [card] [\[T\] Tools & MCP v2 Layer](https://docs.openliate.tech/layers/tools-mcp)
>
> Built-in identity tools (Aadhaar & PAN) and Model Context Protocol (MCP) tool server dispatching.

> [card] [\[E\] Environment & Sandbox Layer](https://docs.openliate.tech/layers/environment-sandbox)
>
> Secret masking (`sk-...`), safe `$VAR` expansion, domain whitelisting, and execution guardrails.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.