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

# Junis Magic

> Manage your entire Junis platform from Claude Desktop, Cursor, or any MCP-compatible client

## What is Junis Magic?

**Junis Magic** is Junis's official MCP (Model Context Protocol) server that gives AI assistants complete access to manage your Junis platform. Instead of navigating the web dashboard, you can build, configure, and orchestrate AI agent teams directly from tools like **Claude Desktop**, **Cursor**, or any MCP-compatible client.

> Build and manage entire AI teams with natural language — no code, no dashboard clicking.

<Info>
  **One-Line Setup**: Just add the Junis Magic MCP server URL and your API key to your MCP client configuration. That's it.
</Info>

***

## Why Junis Magic?

<CardGroup cols={2}>
  <Card title="AI-Powered Management" icon="wand-magic-sparkles">
    Let AI assistants build your agent teams. Describe what you need in natural language, and your assistant creates, configures, and wires everything together.
  </Card>

  <Card title="Complete Platform Control" icon="toolbox">
    A comprehensive toolset across 16 categories — from creating agents to uploading documents, managing schedules, and configuring OAuth integrations.
  </Card>

  <Card title="Cross-Organization" icon="building">
    Manage multiple organizations from a single MCP connection. Switch between teams effortlessly.
  </Card>

  <Card title="Zero Code Required" icon="keyboard">
    No API calls to write. No dashboard to navigate. Just tell your AI what you want.
  </Card>
</CardGroup>

***

## Getting Started

### Step 1: Get Your API Key

1. Log in to [junis.ai](https://junis.ai)
2. Go to **Settings > API Keys**
3. Create a **User API Key** (starts with `jns_user_`)

<Note>
  **User API Keys** are recommended — they work across all your organizations with a single key. You can create up to 5 user API keys.
</Note>

### Step 2: Configure Your MCP Client

<Tabs>
  <Tab title="Claude Desktop">
    Add to your Claude Desktop configuration file:

    **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`

    **Windows**: `%APPDATA%\Claude\claude_desktop_config.json`

    ```json theme={null}
    {
      "mcpServers": {
        "junis-magic": {
          "url": "https://mcp.junis.ai/mcp",
          "transport": "streamable-http",
          "headers": {
            "Authorization": "Bearer jns_user_YOUR_API_KEY"
          }
        }
      }
    }
    ```
  </Tab>

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

    ```json theme={null}
    {
      "mcpServers": {
        "junis-magic": {
          "url": "https://mcp.junis.ai/mcp",
          "transport": "streamable-http",
          "headers": {
            "Authorization": "Bearer jns_user_YOUR_API_KEY"
          }
        }
      }
    }
    ```
  </Tab>

  <Tab title="Other MCP Clients">
    Use these connection details for any MCP-compatible client:

    | Setting         | Value                                         |
    | --------------- | --------------------------------------------- |
    | **URL**         | `https://mcp.junis.ai/mcp`                    |
    | **Transport**   | Streamable HTTP                               |
    | **Auth Header** | `Authorization: Bearer jns_user_YOUR_API_KEY` |
  </Tab>
</Tabs>

### Step 3: Start Building

Once connected, ask your AI assistant to manage your Junis platform:

```
"Switch to my organization and show me all my agents."
"Create a customer support agent with web search enabled."
"Build a research team with a web scraper and a report writer working in sequence."
```

<Note>
  **Build-first**: For clear build requests, your assistant builds immediately with smart defaults and reports afterward — it only asks for confirmation on large builds (10+ agents), destructive changes, or high-cost operations.
</Note>

***

## Available Tool Categories

Junis Magic provides a comprehensive toolset organized in 16 categories:

| Category                    | Key Capabilities                                                  |
| --------------------------- | ----------------------------------------------------------------- |
| **Organization Management** | List, create, and switch between organizations                    |
| **Agent Management**        | Create, update, clone agents (LLM, Sequential, Parallel, Loop)    |
| **Sub-Agent Relationships** | Wire parent-child agent hierarchies, reorder execution            |
| **MCP Platform Management** | Connect and configure MCP integrations for agents                 |
| **RAG Knowledge Base**      | Create datastores, upload documents, track indexing               |
| **Sessions**                | List conversations, view message history, manage sessions         |
| **Chat**                    | Send messages to agents, OpenAI-compatible completions            |
| **Schedules**               | Create and manage automated task schedules                        |
| **Skills**                  | Import prompt-template skills from GitHub or npm                  |
| **Composio OAuth**          | Connect 200+ OAuth integrations (Gmail, Slack, GitHub, etc.)      |
| **Team Builder**            | Batch operations for creating entire agent teams at once          |
| **Content Management**      | Track AI-generated images, videos, and audio                      |
| **Soul & Identity**         | Define your organization's AI personality and communication style |
| **Memory Management**       | View and update user memory profiles                              |
| **RAG Upload**              | Upload and manage knowledge base documents                        |
| **Device Management**       | Connect and manage local devices for desktop agent control        |

***

## Example Workflows

### Build a Complete Agent Team

```
You: "Create a content team with these agents:
1. A research agent that can search the web
2. A writer agent that creates articles
3. A reviewer agent that checks quality
Set them up as a sequential pipeline."

AI Assistant:
1. Creates 3 LLM agents with appropriate instructions
2. Creates a Sequential parent agent
3. Wires sub-agent relationships in order
4. Links web search capability to the research agent

Done! Team ready to use.
```

### Set Up a Knowledge Base

```
You: "Create a knowledge base called 'Product Docs' and upload the product manual."

AI Assistant:
1. Creates a new RAG DataStore
2. Uploads the document
3. Monitors indexing progress
4. Reports when the knowledge base is ready for search
```

### Configure OAuth Integrations

```
You: "Connect Gmail to my support agent so it can read and send emails."

AI Assistant:
1. Lists available Composio integrations
2. Links Gmail integration to the agent
3. Provides the OAuth URL for you to authorize
4. Confirms connection status
```

### Manage Soul & Identity

```
You: "Set our AI personality to be professional but approachable,
with a focus on data-driven recommendations."

AI Assistant:
1. Reads current soul and identity configuration
2. Updates soul.md with reasoning principles
3. Updates identity.md with communication style
4. Changes take effect on next conversation
```

***

## Team Builder: Batch Operations

For complex setups, the **Team Builder** tool lets you create entire agent architectures in a single call:

```
You: "Build me an e-commerce support team with:
- Product FAQ agent (with RAG knowledge base)
- Order tracking agent (with Composio for CRM)
- Escalation agent (routes complex issues)
All orchestrated through a main support agent."

AI Assistant uses apply_team_changes() to:
1. Create all 4 agents simultaneously
2. Configure each with appropriate models and instructions
3. Wire the orchestration hierarchy
4. Link MCP platforms and integrations
```

<Tip>
  **Pro Tip**: The Team Builder is the fastest way to set up complex agent architectures. It handles all the wiring automatically.
</Tip>

***

## Supported Models

Junis Magic lets you configure agents with various AI models:

| Model                                 | Best For                                                                  |
| ------------------------------------- | ------------------------------------------------------------------------- |
| **Claude Sonnet 4.6**                 | Default — general purpose, strong tool calling, balanced cost/performance |
| **Claude Opus 4.8**                   | Flagship — complex reasoning, advanced analysis                           |
| **Gemini 3.5 Flash / 3.1 Flash Lite** | Cost-effective, fast; Gemini is **required** for RAG agents               |
| **GPT-5.x**                           | Alternative general purpose (e.g. GPT-5.5, GPT-5.4)                       |

<Warning>
  **RAG Agents**: Must use Gemini models (Vertex AI Search requirement). Other models cannot use RAG tools.
</Warning>

<Note>
  Models are validated against the platform's live model registry — if you request an unavailable model, your assistant will suggest the closest supported one. Ask *"what models can I use?"* for the current list.
</Note>

***

## Best Practices

<Check>
  **Recommended**:

  * Always call `switch_organization` first when starting a session
  * Use `list_agents` and `list_tools` to understand your current setup before making changes
  * Use the Team Builder for creating multiple agents at once
  * Use User API Keys for cross-organization management
  * Test agents with `send_message` after creation
</Check>

<Warning>
  **Important Notes**:

  * Agent names should use lowercase with hyphens (e.g., `my-agent`)
  * RAG agents must use Gemini models
  * Changes take effect on the next conversation with the agent
</Warning>

***

## Related Guides

<CardGroup cols={2}>
  <Card title="Agent System" icon="robot" href="/getting-started/agent-system">
    Learn about agent types and orchestration
  </Card>

  <Card title="MCP Integration" icon="plug" href="/guides/mcp/overview">
    Configure MCP platforms
  </Card>

  <Card title="Composio Integration" icon="key" href="/integrations/composio">
    Connect OAuth-based services
  </Card>

  <Card title="API Reference" icon="code" href="/api-reference/authentication">
    Programmatic access via REST API
  </Card>
</CardGroup>

***

<Note>
  **Need Help?** Contact us at [contact@junis.ai](mailto:contact@junis.ai) for assistance with Junis Magic setup and configuration.
</Note>
