Back to Blog
Mate Benyovszky

Deep MCP Integration — AgentPlaybooks meets Clawdbot

How to connect your AgentPlaybooks with Clawdbot, the trending open-source self-hosted AI assistant that supports WhatsApp, Telegram, and full MCP protocol.

Deep MCP Integration: AgentPlaybooks meets Clawdbot

The AI assistant landscape just got a major shake-up. Clawdbot, the open-source, self-hosted personal AI assistant created by Peter Steinberger, is trending across developer communities — and for good reason. It combines the power of Claude, ChatGPT, and Gemini with full Model Context Protocol (MCP) support, WhatsApp/Telegram/Discord integration, and complete local control.

Today, we're excited to announce first-class support for Clawdbot in AgentPlaybooks. This means you can now power your Clawdbot with portable, shareable playbooks — complete with skills, personas, and persistent memory.

What is Clawdbot?

Clawdbot is an AI Gateway that acts as a bridge between popular messaging platforms and large language model APIs. Key features include:

  • 🔐 Self-hosted privacy — Runs on your own hardware (typically a Mac Mini)
  • 📱 Multi-platform messaging — WhatsApp, Telegram, Discord, and more
  • 🧠 Persistent memory — Remembers context across conversations
  • 🤖 Proactive messaging — Can initiate conversations based on triggers
  • 🔧 Full MCP support — Extends capabilities via the Model Context Protocol
  • 🌐 Multi-LLM backend — Choose Claude, ChatGPT, Gemini, or local models

Unlike cloud-based AI assistants, Clawdbot keeps your data local and gives you complete control over the AI's behavior.

Why MCP Changes Everything

The Model Context Protocol is an open standard (developed by Anthropic) that allows AI agents to connect to external tools and data sources. Think of it as a universal plugin system for AI.

With MCP, your Clawdbot can:

  • Access real-time data from external APIs
  • Execute tools and functions
  • Read from persistent memory stores
  • Follow structured skill definitions

AgentPlaybooks provides a fully compliant MCP server endpoint for every playbook — making integration seamless.

Setting Up AgentPlaybooks with Clawdbot

Step 1: Create Your Playbook

First, create a playbook at agentplaybooks.ai/dashboard with:

  • Personas — Define how your assistant should behave
  • Skills — What capabilities it should have
  • Memory — Persistent context it should remember

Step 2: Get Your MCP Endpoint

Every public playbook gets an MCP endpoint:

https://agentplaybooks.ai/api/mcp/YOUR_GUID

Step 3: Configure Clawdbot

Add your AgentPlaybooks MCP server to Clawdbot's configuration. In your config.yaml:

# Clawdbot config.yaml
llm:
  provider: anthropic  # or openai, google
  model: claude-sonnet-4-20250514

mcp_servers:
  - name: agent-playbook
    transport: http
    url: https://agentplaybooks.ai/api/mcp/YOUR_GUID
    description: My custom AI playbook with skills and memory

messaging:
  whatsapp:
    enabled: true
    phone_number: "+1234567890"
  telegram:
    enabled: true
    bot_token: "your_bot_token"

Step 4: Verify Connection

Start Clawdbot and verify the MCP connection:

clawdbot status

# Output should show:
# MCP Servers:
#   ✓ agent-playbook (connected)
#     - Tools: 5 available
#     - Resources: 3 available

What Your Clawdbot Gets

Once connected, Clawdbot inherits everything from your playbook:

Tools (from Skills)

Your skills become callable tools. For example:

| Playbook Skill | Clawdbot Tool | |---------------|---------------| | search_web | search_web(query) | | summarize_text | summarize_text(text, length) | | translate_document | translate_document(text, target_lang) |

Resources

Clawdbot can read your playbook resources:

playbook://YOUR_GUID/personas  → AI personalities
playbook://YOUR_GUID/memory    → Persistent storage
playbook://YOUR_GUID/skills    → Capability definitions

Persistent Memory

The killer feature: shared memory across platforms. Update a memory entry via Clawdbot on WhatsApp, and it's immediately available when you chat on Telegram or Discord.

// Clawdbot automatically syncs memory
await mcp.writeResource("playbook://GUID/memory/user_preferences", {
  timezone: "Europe/Budapest",
  language: "hu",
  notification_style: "brief"
});

Real-World Use Cases

🏠 Smart Home Assistant

Use Clawdbot + AgentPlaybooks to create a personal home assistant:

Playbook skills:

  • control_lights — Interface with Home Assistant
  • set_thermostat — Adjust temperature
  • check_calendar — Read today's events
  • send_reminder — Push notifications

Messaging platforms:

  • WhatsApp for family members
  • Telegram for quick commands
  • Discord for home automation logs

📊 Business Intelligence Bot

Playbook skills:

  • query_database — Run SQL queries safely
  • generate_chart — Create visualizations
  • summarize_report — Condense long documents
  • schedule_meeting — Book calendar slots

Playbook memory:

  • Recent queries and results
  • User preferences
  • Commonly used metrics

🌍 Multilingual Support Agent

Playbook personas:

  • Hungarian support persona
  • English support persona
  • German support persona

Skills:

  • detect_language — Auto-detect user language
  • translate_response — Translate before sending
  • log_ticket — Create support tickets

Advanced: Bidirectional Memory Sync

AgentPlaybooks supports write-back via API. Configure Clawdbot to persist memories:

mcp_servers:
  - name: agent-playbook
    transport: http
    url: https://agentplaybooks.ai/api/mcp/YOUR_GUID
    auth:
      type: bearer
      token: apb_live_xxx  # Your AgentPlaybooks API key
    write_enabled: true

Now Clawdbot can update your playbook's memory:

# Inside Clawdbot tool execution
def remember_user_preference(key: str, value: any):
    requests.put(
        f"https://agentplaybooks.ai/api/playbooks/{GUID}/memory/{key}",
        headers={"Authorization": f"Bearer {API_KEY}"},
        json={"value": value}
    )

Sample Clawdbot-Ready Playbook

Here's a complete playbook JSON you can import:

{
  "name": "Clawdbot Personal Assistant",
  "description": "Multi-platform AI assistant with MCP integration",
  "personas": [
    {
      "name": "Assistant",
      "system_prompt": "You are a helpful personal assistant. You can search the web, manage tasks, and remember user preferences. Always be concise in messaging contexts."
    }
  ],
  "skills": [
    {
      "name": "quick_search",
      "description": "Search the web and return a brief summary suitable for messaging",
      "input_schema": {
        "type": "object",
        "properties": {
          "query": {"type": "string"},
          "max_results": {"type": "integer", "default": 3}
        },
        "required": ["query"]
      }
    },
    {
      "name": "set_reminder",
      "description": "Create a reminder that will be sent via the current messaging platform",
      "input_schema": {
        "type": "object",
        "properties": {
          "message": {"type": "string"},
          "when": {"type": "string", "description": "Natural language time like 'in 2 hours' or 'tomorrow at 9am'"}
        },
        "required": ["message", "when"]
      }
    },
    {
      "name": "manage_task",
      "description": "Add, complete, or list tasks from the personal task list",
      "input_schema": {
        "type": "object",
        "properties": {
          "action": {"type": "string", "enum": ["add", "complete", "list"]},
          "task": {"type": "string"},
          "priority": {"type": "string", "enum": ["low", "medium", "high"]}
        },
        "required": ["action"]
      }
    }
  ],
  "memory": {
    "user_name": "Friend",
    "preferred_language": "en",
    "notification_hours": "09:00-22:00"
  }
}

The Privacy Advantage

Unlike cloud-only solutions, the Clawdbot + AgentPlaybooks combo offers:

| Feature | Cloud AI | Clawdbot + AgentPlaybooks | |---------|----------|--------------------------| | Data location | Provider's servers | Your local machine | | Message privacy | Logged by vendor | Stays on device | | Customization | Limited | Unlimited | | Offline capability | None | Full (with local LLM) | | Cost | Per-message fees | Fixed infrastructure |

Getting Started Today

  • Install Clawdbot — Follow the Clawdbot documentation
  • Create a playbook — Design your AI at agentplaybooks.ai/dashboard
  • Connect via MCP — Add your playbook endpoint to Clawdbot config
  • Start chatting — Your AI assistant is ready on WhatsApp, Telegram, and Discord

The combination of Clawdbot's self-hosted infrastructure and AgentPlaybooks' portable playbooks creates the ultimate AI assistant stack — private, powerful, and completely under your control.

Ready to build your own? Create your playbook →

Share this article