Is make-it-heavy open source?

Yes — Doriandarko/make-it-heavy is an open-source project tracked on heatdrop.

What language is make-it-heavy written in?

Doriandarko/make-it-heavy is primarily written in Python.

How popular is make-it-heavy?

Doriandarko/make-it-heavy has 1.1k stars on GitHub.

Where can I find make-it-heavy?

Doriandarko/make-it-heavy is on GitHub at https://github.com/Doriandarko/make-it-heavy.

← all repositories

Doriandarko/make-it-heavy

Clone Grok Heavy by committee: four LLMs in parallel

A Python framework that farms out deep-research queries to concurrent agents via OpenRouter, then merges their perspectives into one heavy answer.

★1.1k stars Python Agents LLMOps · Eval

View on GitHub ↗

Not currently ranked — collecting fresh signals.

star history

What it does Make It Heavy is a Python orchestrator that mimics Grok’s “Heavy” deep-research mode by splitting a user query into specialized sub-questions, dispatching them to four parallel LLM agents via OpenRouter, and synthesizing the results into a single comprehensive answer. It also supports a single-agent mode for simpler tasks. The framework auto-discovers tools from a local directory, giving each agent access to web search, file I/O, and calculations.

The interesting bit The project is essentially well-organized glue code between OpenRouter and multiple agent loops, but the clever part is using an LLM to dynamically generate distinct research angles—like “verify facts” versus “find alternative perspectives”—then running them concurrently. The synthesis step is also handled by an LLM, meaning the entire pipeline is model-driven from decomposition to final merge.

Key highlights

Parallel agent execution with configurable agent count and timeouts via config.yaml
Hot-swappable tool system: drop a Python file inheriting from BaseTool into the tools/ folder and it loads automatically
Built entirely on OpenRouter, so you can swap models (GPT-4.1-mini, Claude 3.5 Sonnet, Llama 3.1 70B, etc.) without touching orchestration logic
Single-agent mode available for lightweight tasks that don’t need a committee
Includes a small built-in toolkit: DuckDuckGo search, calculator, file read/write

Caveats

Requires an OpenRouter API key and budget for multiple concurrent model calls; the README warns that your plan must support the concurrency level you configure
The included tools are basic; anything beyond search, math, and file I/O requires writing your own BaseTool subclass
MIT license carries a commercial attribution clause requiring credit for products with 100K+ users, which is an unusual extra condition

Verdict Worth a look if you want to experiment with multi-agent research pipelines without writing your own orchestration layer, but skip it if you need a polished framework with built-in observability—debugging still requires editing orchestrator.py to enable verbose output.

Frequently asked

What is Doriandarko/make-it-heavy?: A Python framework that farms out deep-research queries to concurrent agents via OpenRouter, then merges their perspectives into one heavy answer.
Is make-it-heavy open source?: Yes — Doriandarko/make-it-heavy is an open-source project tracked on heatdrop.
What language is make-it-heavy written in?: Doriandarko/make-it-heavy is primarily written in Python.
How popular is make-it-heavy?: Doriandarko/make-it-heavy has 1.1k stars on GitHub.
Where can I find make-it-heavy?: Doriandarko/make-it-heavy is on GitHub at https://github.com/Doriandarko/make-it-heavy.