Make LLMs from different vendors collaborate as one team. Capability-based task routing across Claude / GPT / Gemini / Ollama. Lead dispatches → broker filter-broadcasts → workers (any LLM, any host) pick up matching tasks. Single-vendor stacks already have subagents — this is for when the right tool for the job lives in another vendor's stack.
-
Updated
May 24, 2026 - Python