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

# Introduction

> Build on the Syvon platform. Manage your account with the portal surface, read published agents and run chat with the brain surface.

Syvon is a multi-tenant platform for AI-generated brand content. A **workspace** owns brands, agents (projects), workflows, files, and a published feed. If you are building a client, an integration, or an agent that talks to Syvon, these docs are your contract.

There are exactly two client-facing surfaces, each with its own credential:

| Surface                                   | What it does                                                                 | Credential                                       |
| ----------------------------------------- | ---------------------------------------------------------------------------- | ------------------------------------------------ |
| **Portal** (`https://syvon.ai/api/*`)     | The management plane. Workspaces, members, brands, API keys, credits, flows. | Portal API JWT (user-scoped, \~1 hour)           |
| **Brain** (`https://brain.syvon.ai/v1/*`) | The read plane. Published agent config, items, feed, files, and public chat. | Workspace API key `sk_ws_...` (workspace-scoped) |

A single credential cannot drive both. That is intentional, not an SDK limitation.

The easiest way in is the typed client SDK:

```bash theme={null}
npm install @syvon/sdk
```

```ts theme={null}
import { SyvonClient } from '@syvon/sdk';

const client = new SyvonClient({ workspaceKey: process.env.SYVON_WS_KEY! });

for await (const ev of client.brain.chat(agentId, {
  prompt: 'Tell me about your work.',
})) {
  if (ev.text) process.stdout.write(ev.text);
}
```

## Who these docs are for

* **Application developers** wiring a Syvon-powered site, feed, or chat into their own product.
* **Agents and LLMs** building on top of Syvon programmatically. Every method lists its HTTP method + path, parameters, return type, required credential, and the server-side behavior that matters to a caller. This site also serves `llms.txt` and `llms-full.txt` at the root for machine reading.
* **Internal teams** documenting what the supported client path is (and what deliberately is not exposed; see [Scope](/scope)).

## What you can build

* A website or app that renders a published agent's feed, files, and site config.
* A chat client that streams an agent's replies (the owner is billed per turn; see [Chat stream](/guides/chat-stream)).
* Account tooling: workspace and member management, credit balance and history, API key minting.