HydraTeams

HydraTeams
The model doesn't matter. The orchestration does.

A translation proxy that lets Claude Code Agent Teams use any AI model as a teammate. GPT Codex, Gemini, Ollama: they all become full Claude Code agents with 15+ tools, file access, bash, git, and autonomous task execution.

One proxy. One env var. Any model.

Status: Working. OpenAI + ChatGPT subscription providers tested end-to-end.

How It Works

Claude Code Agent Teams spawns teammates as separate Claude Code processes. Each teammate communicates with its LLM via the Anthropic Messages API. HydraTeams is a proxy that intercepts these API calls and translates them to any provider's format.

The teammate is still a full Claude Code instance with every tool — Read, Write, Edit, Bash, Glob, Grep, Git. It just doesn't know its brain is GPT instead of Claude.

Why HydraTeams?

You already have the best agent framework. Claude Code Agent Teams is a battle-tested multi-agent system with agentic loops, 15+ tools, file-based coordination, task dependency graphs, messaging, plan approval, and graceful shutdown. Building another one is reinventing the wheel. HydraTeams makes Agent Teams model-agnostic instead.

Real cost savings. Not every task needs a $15/M token frontier model. Route research to Gemini Flash ($0.01), codegen to Codex ($0.12), architecture to Opus ($0.15). Same team, smart routing, real savings. Or use your ChatGPT Plus subscription and pay $0 extra.

Zero vendor lock-in. If OpenAI is down, route through Gemini. If prices change, switch. New model drops? Update one config value.

Every tool, every capability. Unlike lightweight agent wrappers, each teammate is a full Claude Code instance. It reads code, writes code, runs tests, uses git, searches files — everything Claude Code does. The proxy only…