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

# Introduction

> An identity-first, multiplayer runtime for humans and agents.

Mesh is a shared conversational runtime for humans and agents. It keeps track
of who said what, which conversation it belongs to, and which agent owns the
work. One installation can host multiple agents with separate identities,
credentials, memory, and context.

## Start here

* [Philosophy](/philosophy): the principles behind Mesh.
* [Architecture layers](/layers): how the runtime fits together.
* [Perspective](/perspective): how each agent's identity and knowledge stay separate.
* [Agent provisioning](/agent-provisioning): configure an agent and its connections.
* [Operator API](/operator-api): administer an installation programmatically.

## Explore the system

* [Agent harness](/runtime/agent-harness) and [turn pipeline](/runtime/turn-pipeline)
  explain how a message becomes an attributed, auditable agent turn.
* [Memory retrieval](/runtime/memory-retrieval) explains how agents recall prior context.
* [Tools](/runtime/tools), [remote MCP](/runtime/remote-mcp), and
  [sandboxed execution](/runtime/sandboxed-execution) cover external capabilities.
* [Connectors](/connectors/generic) connect the runtime to communication platforms.
* [Data model](/data-model) and [persistence](/persistence) describe durable state.

These pages include architecture contracts and design proposals as well as
operating instructions. A proposed capability is not a claim that it has shipped;
check each page's implementation status and the
[repository](https://github.com/TextureHQ/mesh) for the current code.


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