How does the skill manage API costs?

It includes built-in rate limiting, max consultation caps per hour, and the ability to toggle specific agents off to ensure API usage remains within your designated budget.

How does the 'Veto Authority' mechanism work?

The Risk Guardian agent has the power to independently block any trade or action that violates safety limits or risk parameters, providing a critical safety layer that can override the orchestrator.

Is this skill safe for live trading environments?

Yes, it is designed with a 'Safety First' approach, offering Advisory and Supervised modes that require human-in-the-loop validation before transitioning to fully autonomous execution.

Can I use this with any trading platform?

While the reference implementation uses Alpaca, the multi-agent logic is designed as a wrapper pattern that can be integrated into most Python-based trading frameworks.

What models are used in the multi-agent system?

The system uses a tiered approach: Claude Opus acts as the Unified Orchestrator, Sonnet handles complex analysis and risk auditing, and Haiku manages high-frequency monitoring and scaling tasks.

Multi-Agent Trading Integration

Name: Multi-Agent Trading Integration
Author: smith6jt-cop

bysmith6jt-cop

データサイエンスとML

Integrates multi-agent Claude systems into algorithmic trading pipelines to optimize model training and manage live risk with automated oversight.

概要

This skill implements a sophisticated multi-agent architecture using Claude Opus, Sonnet, and Haiku to oversee financial trading workflows. It coordinates specialized agents across three domains: training optimization (hyperparameters, rewards), live trading (signal analysis, risk guarding), and portfolio selection. By utilizing an orchestrator-subagent pattern, it provides a robust safety layer with veto authority, strict operational bounds, and rate limiting, making it an essential tool for developers building AI-assisted trading systems that require high-reliability decision-making.

主な機能

Multi-tiered architecture involving Training, Trading, and Selection agents
0 GitHub stars
Real-time Risk Guardian with veto authority over live financial execution
Automated hyperparameter tuning and reward engineering for RL models
Comprehensive safety guardrails with numeric clamping and rate limiting
Configurable operating modes: Advisory, Supervised, and Autonomous

ユースケース

Orchestrating ensemble LLM feedback to validate market signals and portfolio diversity
Implementing an AI-powered risk management layer for live trade execution and position sizing
Optimizing reinforcement learning hyperparameters during algorithmic trading model training

概要

主な機能

Multi-tiered architecture involving Training, Trading, and Selection agents
0 GitHub stars
Real-time Risk Guardian with veto authority over live financial execution
Automated hyperparameter tuning and reward engineering for RL models
Comprehensive safety guardrails with numeric clamping and rate limiting
Configurable operating modes: Advisory, Supervised, and Autonomous

ユースケース

Orchestrating ensemble LLM feedback to validate market signals and portfolio diversity
Implementing an AI-powered risk management layer for live trade execution and position sizing
Optimizing reinforcement learning hyperparameters during algorithmic trading model training