Agent System Reference

This document provides comprehensive technical documentation for the agent swarm system within the Signal trading platform. The agent system consists of specialized AI consultants that provide cryptocurrency trading analysis and recommendations through a coordinated multi-agent architecture.

For information about the core AI trading engine components (SignalOutline, CloseOutline, RiskOutline), see AI Trading Engine. For configuration of agent behavior and AI model selection, see Environment Configuration.

The agent swarm system implements a triage-based routing architecture where user queries are distributed to specialized cryptocurrency consultant agents. Each agent operates with dedicated analysis tools and maintains expertise in specific digital assets while sharing common infrastructure for market data access and AI inference.

Mermaid Diagram

All agents in the system share a common architectural foundation while maintaining specialized functionality. The following diagram maps the natural language agent descriptions to their actual code entities and configuration structures.

Mermaid Diagram

All agents share the following architectural characteristics:

Property Value Description
Completion Engine runner_stream_completion Streaming AI completion system
Language Mode Всегда пиши ответ на русском языке Russian-language responses
Reasoning Level Reasoning: high High-level analytical reasoning
Error Handling onToolError callback Standardized error management
Message Format Markdown text only No HTML, files, or tables supported

The agent tool system provides standardized interfaces for market data access across different timeframes and analysis types. Each tool category serves specific analytical purposes while maintaining consistent parameter structures.

Mermaid Diagram

Tool Category Purpose Data Scope Update Frequency
Signal Analysis Primary technical analysis 36-48 hour lookback Real-time
Historical Data Trend analysis over time 5-30 stored calculations 2-5 minutes
Market Structure Liquidity and order flow 2-220 hour windows Real-time
Social Sentiment Community sentiment analysis Recent posts On-demand
Raw Candles OHLCV price action 6-15 recent candles Real-time

The agent system implements a hub-and-spoke navigation model with the triage_agent serving as the central router. Navigation between agents occurs through dedicated tools that enable seamless conversation handoffs.

Mermaid Diagram

The triage_agent implements keyword-based routing logic to determine appropriate consultant assignment:

Cryptocurrency Agent Target Trigger Keywords Navigation Tool
Bitcoin btc_consultant_agent BTC, Bitcoin, Биткоин navigate_to_btc_consultant_tool
Ethereum eth_consultant_agent ETH, Ethereum, Эфириум, Эфир navigate_to_eth_consultant_tool
Binance Coin bnb_consultant_agent BNB, Binance Coin, Binance Smart Chain navigate_to_bnb_consultant_tool
Ripple xrp_consultant_agent XRP, Ripple, Рипл navigate_to_xrp_consultant_tool
Solana sol_consultant_agent SOL, Solana, Солана navigate_to_sol_consultant_tool

The triage_agent serves as the central routing hub for the agent swarm system. It analyzes user queries to determine cryptocurrency interest and routes conversations to appropriate specialist agents.

Core Functionality:

  • Conversation topic detection and routing
  • Natural language interaction without explicit agent switching
  • Support for general conversation when topics fall outside cryptocurrency scope
  • Seamless handoff to specialist consultants

System Behavior:

  • Responds in Russian language
  • Maintains conversational flow without explicit greetings
  • Routes immediately upon cryptocurrency mention detection
  • Does not attempt extended conversation on off-topic subjects

The btc_consultant_agent provides specialized Bitcoin trading analysis and consultation services with comprehensive technical analysis capabilities.

Market Context:

  • Current Bitcoin price tracking: 111983.60
  • Date/time awareness: 08/09/2025 17:14 +05:00
  • Specialized Bitcoin analysis tools and indicators

Tool Arsenal:

  • Complete signal analysis suite (15min, 30min, 1hour, 1min timeframes)
  • Historical indicator tracking with 5-30 calculation history
  • Social media sentiment analysis (Mastodon, Twitter)
  • Advanced market structure analysis (volume, whale detection, order book)
  • Real-time candle data across multiple timeframes

Navigation Behavior:

  • Returns to triage_agent when user interest shifts to other cryptocurrencies
  • Seamless conversation continuity without explicit handoff announcements
  • Immediate tool invocation based on user query requirements

The eth_consultant_agent provides specialized Ethereum trading analysis with identical tool capabilities tailored for ETH market analysis.

Market Context:

  • Current Ethereum price tracking: 4307.81
  • Complete technical analysis coverage for Ethereum markets
  • Ethereum-specific social sentiment monitoring

Specialized Features:

  • Same comprehensive tool suite as Bitcoin consultant but Ethereum-focused
  • Ethereum market pattern recognition and analysis
  • ETH-specific whale activity detection and institutional flow analysis

The bnb_consultant_agent specializes in Binance Coin (BNB) trading analysis and Binance Smart Chain ecosystem consultation.

Market Context:

  • Current BNB price tracking: 877.69
  • Binance Smart Chain ecosystem analysis capabilities
  • BNB-specific market structure analysis

Tool Configuration:

  • Complete technical analysis suite for BNB markets
  • Binance ecosystem-specific insights and analysis
  • BSC network activity and token flow analysis capabilities

The sol_consultant_agent provides specialized Solana (SOL) trading analysis and ecosystem consultation services.

Market Context:

  • Current SOL price tracking: 214.08
  • Solana ecosystem analysis and market intelligence
  • SOL-specific technical analysis and pattern recognition

Analysis Capabilities:

  • Complete signal analysis across all timeframes for Solana
  • Solana network activity and validator performance insights
  • SOL market microstructure and liquidity analysis

All consultant agents maintain consistent behavioral patterns:

  1. Language Consistency: All responses in Russian language
  2. Tool Efficiency: Tools invoked only when required by user queries
  3. Navigation Transparency: Agent switching occurs without user notification
  4. Content Restrictions: Text-only markdown responses, no file generation
  5. Price Integration: Current price inclusion when making predictions
  6. Social Sentiment: Link inclusion and sentiment classification for social media queries