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

# Agent Quickstart

> Establish an MCP connection and verify it with a safe, read-only call.

This guide gets you from zero to a working MCP connection with one safe,
read-only verification call. It does not create or approve work.

## Step 1: Connect

<Tabs>
  <Tab title="OpenClaw (Recommended)">
    Install the documented plugin, then complete browser pairing:

    ```bash theme={"dark"}
    openclaw plugins install @useorgx/openclaw-plugin
    ```

    Open the plugin's local dashboard, choose **Connect OrgX**, and approve in
    the browser.

    [Full OpenClaw setup guide →](/docs/guides/openclaw-plugin-setup)
  </Tab>

  <Tab title="Cursor">
    Add to `~/.cursor/mcp.json`:

    ```json theme={"dark"}
    {
      "mcpServers": {
        "orgx": {
          "command": "npx",
          "args": ["mcp-remote", "https://mcp.useorgx.com/mcp"]
        }
      }
    }
    ```

    Restart Cursor. On first use, `mcp-remote` opens your browser for OAuth sign-in automatically.
  </Tab>

  <Tab title="Claude">
    In Claude or Claude Desktop, open **Settings → Connectors** and add a
    custom remote MCP connector using:

    ```text theme={"dark"}
    https://mcp.useorgx.com/mcp
    ```

    Then complete the OAuth flow in your browser.
  </Tab>
</Tabs>

<Tip>
  Need client-specific instructions for ChatGPT, VS Code, or other MCP hosts?
  Use the full [MCP Client Setup guide](/docs/guides/cursor-mcp-setup).
</Tip>

## Step 2: Verify Connection

Run this safe, read-only call to confirm everything works:

```json theme={"dark"}
{
  "tool": "workspace",
  "args": { "action": "get" }
}
```

The response should identify the active workspace. IDs, names, and additional
fields vary by account; do not copy a response from this page as workspace
state.

If you get `workspace_not_set`, list available workspaces and set one:

```json theme={"dark"}
{ "tool": "workspace", "args": { "action": "list" } }
```

```json theme={"dark"}
{
  "tool": "workspace",
  "args": { "action": "set", "workspace_id": "<workspace_id>" }
}
```

## Step 3: Explore Your Org

```json theme={"dark"}
{
  "tool": "orgx_recommend",
  "args": { "mode": "morning_brief", "period": "30d" }
}
```

The response contains the available brief or recommendation signals for the
active workspace. Its contents depend on the workspace and reporting period.

## Step 4: Check Pending Decisions

```json theme={"dark"}
{
  "tool": "orgx_search",
  "args": {
    "type": "decision",
    "status": "pending",
    "limit": 5
  }
}
```

The response contains the matching decisions, if any. Review the returned
record before taking a write action.

If you have pending decisions, approve one:

```json theme={"dark"}
{
  "tool": "orgx_decide",
  "args": {
    "action": "approve",
    "decision_id": "<decision_id>",
    "note": "Approved after reviewing the artifact and evidence"
  }
}
```

## What's Next

<CardGroup cols={2}>
  <Card title="Agent Recipes" icon="book" href="/docs/agent-ops/agent-recipes">
    Small, reviewable workflows with contract links.
  </Card>

  <Card title="Failure Playbooks" icon="shield" href="/docs/agent-ops/failure-playbooks">
    Error recovery for common issues.
  </Card>

  <Card title="Tool Profiles" icon="users" href="/docs/agent-ops/tool-profiles">
    Reduce schema overhead with role-based profiles.
  </Card>

  <Card title="Full Tool Reference" icon="wrench" href="/docs/api/mcp-tools">
    Complete documentation for the current MCP tool catalog.
  </Card>
</CardGroup>