Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.agentbot.raveculture.xyz/llms.txt

Use this file to discover all available pages before exploring further.

Agents are autonomous AI assistants that can interact with users across multiple platforms. Agentbot agents

​
How Agents Work

Each agent runs in an isolated Docker container with:
  • AI Model - Powered by OpenRouter by default
  • Memory - Persistent conversation history
  • Tools - API integrations and capabilities
  • Personality - Custom instructions and behavior

​
Agent Structure

interface Agent {
  id: string;
  name: string;
  description: string;
  
  // AI Configuration
  model: string;           // e.g., "anthropic/claude-3-opus"
  temperature: number;     // 0-1, creativity level
  maxTokens: number;
  
  // System Prompt
  instructions: string;    // Agent personality & behavior
  
  // Capabilities
  tools: Tool[];
  integrations: Integration[];
  
  // Memory
  memoryEnabled: boolean;
  memoryLimit: number;     // Max messages to remember
}

​
Creating an Agent

​
Via Dashboard

  1. Go to Dashboard → New Agent
  2. Choose a template or blank agent
  3. Configure:
    • Name and description
    • AI model and settings
    • System instructions
    • Enabled tools
  4. Deploy

​
Via API

curl -X POST https://agentbot.sh/api/agents \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "My Agent",
    "model": "anthropic/claude-3-opus",
    "instructions": "You are a helpful assistant...",
    "tools": ["web-search", "calculator"]
  }'

​
Agent Templates

Agentbot includes pre-built templates:
TemplateDescription
AssistantGeneral purpose AI assistant
Customer SupportSupport bot with knowledge base
Rave EventEvent management & guest lists
TreasuryCommunity fund management

​
Tools

Agents can use tools to extend their capabilities:
  • Web Search - Search the internet
  • Calculator - Math operations
  • Weather - Get weather data
  • Custom APIs - Your own API endpoints

​
Feedback and corrections

You can submit corrections to teach an agent what it did wrong and what it should do instead. The agent stores these corrections in memory and uses them to improve future responses. Feedback entries are categorized by type — tone, accuracy, format, behavior, or general — so the agent knows which aspect of its behavior to adjust. See the Feedback API for endpoint details and examples.

​
Persistence

Agents maintain memory across conversations: 
// Configure memory
const agent = {
  memoryEnabled: true,
  memoryLimit: 100,  // Keep last 100 messages
  // Older messages are summarized and stored
}

​
Fleet monitoring

You can monitor your agent fleet from the mission control dashboard. Fleet data — including execution traces, cost breakdowns, and talent bookings — is sourced from treasury transaction records. This gives you a unified view of what your agents are doing and how much they cost.
  • Traces — the 50 most recent agent actions, including coordination messages and AI inference costs
  • Costs — spending grouped by agent and category (for example ai_metric or agent_message)
  • Bookings — talent booking records created through agent-to-agent negotiation, with full pricing lifecycle
See the Mission Control API reference for endpoint details. To track token consumption and spending trends over time, use the dashboard cost API.

​
Best practices

  1. Clear Instructions - Write specific system prompts
  2. Limited Tools - Only enable necessary tools
  3. Memory Management - Set appropriate memory limits
  4. Monitor Costs - Track API usage in dashboard