Skip to content
Dhanasvi
Sign inExplore
GenAgent logo

GenAgent

Verified

LLM-powered agent for building collaborative AI workflows inside ComfyUI.

Autonomous AgentsAutomation 203Open source
View on GitHub
Updated 2026-06-15
GenAgent GitHub repository

What is GenAgent?

GenAgent is an open-source implementation centered on ComfyUI that turns task instructions into executable visual workflows representing multi-component AI systems. It supplies both a benchmark suite and an agent that can study existing examples before proposing fresh designs.

The agent reads node documentation, studies curriculum workflows, and outputs new graphs that are translated into code for easier LLM reasoning. Execution success and requirement fulfillment are tracked automatically when a ComfyUI server is available.

It is intended for researchers and developers who want to measure or improve how language-model agents compose complex, collaborative pipelines without manual graph construction.

Capabilities

generate comfyui workflows
design collaborative ai systems
learn from existing workflows
autonomously execute tasks
benchmark agent performance

What you can build with GenAgent

Autonomous pipeline creation

Generate complete ComfyUI workflows from natural-language task descriptions for image, video or data processing.

Agent benchmarking

Run standardized tests on 200 tasks to compare pass rate and resolve rate across different models or prompting strategies.

Workflow learning and reuse

Study the 20 provided curriculum workflows and documentation for 3205 nodes to improve future designs.

Install GenAgent

Install
git clone https://github.com/xxyQwQ/ComfyBench && cd ComfyBench && conda create -n comfybench python=3.12 && pip install -r requirements.txt
Quick start
git clone https://github.com/xxyQwQ/ComfyBench
cd ComfyBench
  1. 1Clone the repository and move into the project folder.
  2. 2Create a conda environment with Python 3.12 and install the listed dependencies.
  3. 3Edit config.yaml to point to your ComfyUI server and supply any required API keys.
  4. 4Place necessary models and extensions inside ComfyUI as described in the repository.
  5. 5Run the main script with an instruction string and agent name to produce and execute a workflow.

Works with

ComfyUIPython

GenAgent: pros & cons

Pros

  • +Offers a ready-to-use benchmark with 200 tasks and clear success metrics.
  • +Converts visual workflows into code that LLMs can read and edit more easily.
  • +Includes 20 example workflows and full node documentation for rapid learning.
  • +Codebase released under an open-source license with CVPR 2025 acceptance.

Cons

  • –Still requires manual download and placement of many ComfyUI models and extensions.
  • –Only functions inside the ComfyUI ecosystem and needs a running server.
  • –Evaluation depends on correct environment setup and can be sensitive to model choice.
Did you find this helpful?

Frequently asked questions

It produces ComfyUI workflow files that describe collaborative AI systems and can be executed automatically.

User reviews

Verified reviews from the community shape this listing's rating.

Loading reviews…

Sign in to review

Promote GenAgent

Add this badge to your website, or share the tool.

DFeatured on DhanasviGenAgent 0