# I built BHive: a multi-agent framework for Gemini/OpenAI/Anthropic/Ollama. Tear it apart.

> Source: <https://dev.to/dcooper/i-built-bhive-a-multi-agent-framework-for-geminiopenaianthropicollama-tear-it-apart-2c8c>
> Published: 2026-09-22 19:33:28+00:00

I've been building this for two weeks and want honest technical feedback before I invest more time.

Bhive is a multi-agent framework in TypeScript. Agents ("Bees") can delegate tasks to each other, call tools, and it works with any LLM provider — Gemini, OpenAI, Anthropic, or local Ollama — through the same code.

`LLMAdapter` interface, with a `ToolCallingParser` that normalizes tool-call formats across providers. Good pattern, or overengineered for what it does?`ToolCallValidator`: injection guards, allowlists, size limits) — anything missing?`createDelegationTool`) makes sense as an abstraction
Repo: [https://github.com/dcooperdev/bhive-core](https://github.com/dcooperdev/bhive-core)

Tests: 299 passing, ~98% coverage. But test coverage isn't the same as "this is well-designed" — that's what I actually want checked.