Getting Started

This guide walks you through from registration to your first Bot conversation — about 10 minutes.

Overview

Monstrum is an AI Agent governance platform. You can create Bots on it, bind tools and resources to them, then let them execute tasks through conversation or API. The platform handles permission control, credential isolation, and operation auditing, ensuring Bots only do what they’re authorized to do.

This guide covers the following steps:

  1. Register an account and create a workspace
  2. Configure an LLM provider
  3. Create your first Bot
  4. Chat with the Bot
  5. (Optional) Bind resources to the Bot

Step 1: Register and Log In

Visit the Monstrum platform and click “Sign Up” to create an account. You’ll need to provide:

  • Email: Used for login and verification emails
  • Username: 3-20 characters, supporting letters, numbers, underscores, and hyphens
  • Password: At least 8 characters
  • Invite Code: If the platform has invite-only registration enabled

After successful registration, the platform automatically creates a personal workspace named after your username.

If email verification is required, check your inbox and click the verification link.

Step 2: Configure an LLM Provider

Bots need an LLM for reasoning. After logging in, you need to configure at least one LLM provider.

  1. Click LLM Providers in the left navigation bar
  2. Click Add Provider
  3. Select the provider type (e.g., OpenAI, Anthropic, Azure, etc.)
  4. Enter your API Key
  5. If you need a custom API address, enter the Base URL (e.g., when using an API proxy)
  6. Click Test Connection to confirm the configuration is correct
  7. Save

After saving, the platform automatically fetches the list of supported models from that provider. You can click Set as Default to make a provider the workspace default.

If this is the first provider, the platform will automatically set it as the default.

Step 3: Create Your First Bot

  1. Click Bot Management in the left navigation bar
  2. Click Create Bot
  3. Fill in the basic information:
    • Bot Name: Give your Bot a name, such as “Assistant”
    • Description (optional): Describe the Bot’s responsibilities
  4. Click Create

After creation, you’ll enter the Bot detail page. At this point, the Bot uses the workspace default LLM provider and model.

Basic Configuration

In the Settings tab of the Bot detail page, you can adjust:

  • LLM Provider: Select a specific provider and model, or use the workspace default
  • System Prompt: Customize the Bot’s identity and behavioral instructions (leave blank to use the default template)
  • Temperature: Control output randomness (0 = deterministic, 1 = creative)
  • Max Iterations: Maximum reasoning rounds per conversation
  • Timeout: Timeout in seconds for a single task
  • Monthly Budget: Monthly cap for Token consumption (USD)

The default configuration works fine for now.

Step 4: Chat with the Bot

  1. On the Bot detail page, switch to the Overview tab
  2. In the chat box at the bottom, type a message like “Hello, introduce yourself”
  3. Press Enter to send

The Bot will use the configured LLM for reasoning and reply. The conversation supports streaming output — you can see the Bot generating its response token by token.

Chat Interface Features

  • New Conversation: Start a new session, clearing the previous context
  • End Session: Manually end the current session, triggering memory extraction
  • Tool Call Display: If the Bot calls a tool, the interface shows the tool name and execution result

At this point, the Bot has no resources bound, so it can only have plain text conversations. Next, you can bind resources to give it the ability to interact with external systems.

Step 5 (Optional): Bind Resources

After binding resources to a Bot, it can call the corresponding tools. Here are the most common scenarios.

Scenario 1: Let the Bot Search the Web

  1. Click Resource Management in the left navigation bar
  2. In the “Available Resources” area, find Web Access and click to enter
  3. Fill in the resource name (e.g., “Web Search”)
  4. Select a search engine — DuckDuckGo requires no API Key and works out of the box
  5. Save the resource
  6. Go back to the Bot detail page → Resource Config tab
  7. Click Bind Resource → Select the Web Access resource you just created
  8. Configure permissions (you can allow all operations, or restrict to search-only/fetch-only)
  9. Save the binding

Now go back to the chat interface and try asking the Bot “search for today’s news” — you’ll see it call the web_search tool to fetch search results.

Scenario 2: Let the Bot Access a Remote Server

  1. In Resource Management, add an SSH type resource
  2. Configure the host address, port, and username
  3. Add credentials (SSH Key or password)
  4. Bind the resource to the Bot on its detail page, and set permission constraints:
    • Host whitelist: Restrict accessible hosts (e.g., 10.0.1.*)
    • Command whitelist: Restrict executable commands (e.g., ls *, cat *)
  5. Save the binding

The Bot can now execute authorized commands on the remote server via SSH.

Scenario 3: Connect an MCP Tool Server

  1. In Resource Management, add an MCP Endpoint type resource
  2. Enter the MCP server address (Streamable HTTP endpoint)
  3. If authentication is needed, configure API Key or OAuth
  4. After saving, the platform automatically discovers available tools on the MCP server
  5. Bind the resource to the Bot on its detail page, and select which discovered tools the Bot is allowed to use

Next Steps

Congratulations! You’ve completed the basic platform setup and your first Bot conversation. Next, you can:

FAQ

Bot replies “No LLM provider configured”

You haven’t configured an LLM provider yet. Go to the LLM Providers page and add at least one provider.

Bot doesn’t call any tools

The Bot has no resources bound, or the bound resource permissions don’t include that operation. Check the Resource Config tab on the Bot detail page.

No response from the conversation

Check the following:

  • Whether the LLM provider’s API Key is valid (click “Test Connection” on the provider page)
  • Whether the Bot is in a “Paused” state
  • Whether the monthly budget has been exhausted

How to reset Bot state

Click New Conversation on the Bot detail page to start a fresh session. If you need to clear all Bot memories, go to the Memory tab.

Edit this page on GitHub