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

# OutX for AI Agents

> Integrate OutX with AI agents, LLMs, and developer tools. Access docs as markdown, use the skill file, and connect via MCP.

OutX is built for AI agent integration. Use these resources to connect your AI workflows to LinkedIn data and social listening.

## Get Your API Key

Before integrating, get your API key:

1. Visit [mentions.outx.ai/api-doc](https://mentions.outx.ai/api-doc) and click **"Reveal API Key"**
2. Store it as an environment variable: `export OUTX_API_KEY="your-key"`

Or authenticate programmatically via [OTP](/linkedin-api/auth-send-otp), no browser needed.

## Docs for Agents

Four ways to connect OutX to your AI agent:

<CardGroup cols={2}>
  <Card title="MCP Server" icon="plug" href="/integrations/mcp">
    Connect Claude, Cursor, or any MCP-compatible agent to OutX. 20 tools covering the full API. Best for interactive agents.
  </Card>

  <Card title="Skill File" icon="brain" href="https://outx.ai/docs/outx-skill.md">
    Structured API reference with guardrails, parameter tables, and code examples. Best for system prompts.
  </Card>

  <Card title="llms.txt" icon="file-lines" href="https://outx.ai/docs/llms.txt">
    Concise index of all documentation pages with one-line descriptions. Best for discovery.
  </Card>

  <Card title="Full Docs" icon="book" href="https://outx.ai/docs/llms-full.txt">
    Complete documentation concatenated into a single file. Best for comprehensive context.
  </Card>
</CardGroup>

## Quick Start Prompts

Copy these prompts into your AI agent to get started:

### Monitor LinkedIn for keywords

```text theme={null}
Use the OutX API (base URL: https://api.outx.ai, auth: x-api-key header) to:
1. Create a keyword watchlist: POST /api-keyword-watchlist with {"keywords": ["your keyword"], "fetchFreqInHours": 6}
2. After the next fetch cycle, retrieve posts: GET /api-posts?watchlist_id=WATCHLIST_ID&sort_by=recent_first
3. Like relevant posts: POST /api-like with {"post_id": "ID_FROM_POSTS", "user_email": "you@company.com"}
```

### Fetch a LinkedIn profile

```text theme={null}
Use the OutX API (base URL: https://api.outx.ai, auth: x-api-key header) to:
1. Fetch profile: POST /linkedin-agent/fetch-profile with {"profile_slug": "linkedin-slug"}
2. Poll: GET /linkedin-agent/get-task-status?api_agent_task_id=TASK_ID (every 5s until status is "completed")
3. Read profile data from task_output
Note: All LinkedIn Data endpoints are async, always poll for results.
```

## AI Builder Integrations

### MCP (Claude Desktop / Cursor / Claude Code)

Install the OutX MCP server for full API access via natural language:

```bash theme={null}
npx outx-mcp-server
```

See [MCP Server setup guide](/integrations/mcp) for configuration details.

### Skill File (System Prompts)

Add the OutX skill file to your project for static API context:

```bash theme={null}
curl -o outx-skill.md https://outx.ai/docs/outx-skill.md
```

Reference it in your `CLAUDE.md` or Cursor rules to give the AI agent full API context.

### Any LLM / Agent Framework

Include the skill file content in your system prompt, or point your agent to `https://outx.ai/docs/llms.txt` for documentation discovery. See also the [LangChain](/integrations/langchain) and [Python SDK](/integrations/python-sdk) integrations.

## What's Next

<CardGroup cols={2}>
  <Card title="Watchlist Quick Start" icon="rocket" href="/api-reference/quickstart">
    Create watchlists and retrieve posts via API
  </Card>

  <Card title="LinkedIn Data Quick Start" icon="bolt" href="/linkedin-api/quickstart">
    Fetch LinkedIn profiles in 2 minutes
  </Card>

  <Card title="API Reference" icon="code" href="/api-reference/introduction">
    Full API documentation
  </Card>

  <Card title="MCP Server" icon="plug" href="/integrations/mcp">
    Connect AI agents to OutX via MCP
  </Card>
</CardGroup>