Building a Policy-Governed Multi-Agent Financial Research Workflow with Omnigent

A tutorial shows how to build a policy-governed multi-agent financial research workflow with Omnigent in an isolated Python environment.

It defines reusable Python tools for live FX lookup and word counting, then configures a lead research agent and a text-auditing sub-agent in YAML.

The workflow runs through the Claude Agent SDK with environment-based Anthropic API key handling and non-interactive policy controls.

The demo fetches a live USD-to-EUR rate, drafts a client-facing summary, and checks the output for quality and compliance.

It highlights how to use live market data while limiting calls, costs, and output risk.