
AgentForge
Open-source Python framework for building multi-agent systems with YAML-defined agents
- Pricing
- Open source
- Platforms
- CLI, self-hosted, Linux
- GitHub stars
- 864
- Last commit
- 3 months ago
- Language
- Python
- License
- GPL-3.0
Similar agents
About AgentForge
AgentForge is an open-source agent framework for multi-agent frameworks. It is available on CLI, self-hosted, Linux, macOS and Windows.
The README presents AgentForge as a low-code environment for prototyping, testing and iterating on autonomous agents and cognitive architectures. Its building blocks are Agents, declarative Cogs that wire agents and memory into workflows, and optional Memory nodes that Cogs declare and pass to agents.
Agents, prompt templates and personas are configured through YAML files, and prompts can be edited while the system is running without a restart. The project says it works with OpenAI, Google Gemini and Anthropic Claude models, as well as local models served through Ollama or LM Studio, and that different agents in one setup can use different models.
The framework suits Python developers who want to experiment with multi-agent orchestration and branching workflows without writing all the glue code themselves. The README points new users to a docs hub with a quickstart and a first real model run guide.
What it does
- Declarative Cogs
- Compose agents into multi-step workflows with branching logic and memory using YAML files.
- YAML agent configuration
- Define agents, prompt templates and personas in YAML rather than code.
- Integrated memory
- Declare memory nodes in Cogs so they are made available to agents automatically.
- Live prompt editing
- Change prompts while the system is running without restarting it.
- Multiple LLM providers
- Run different agents against OpenAI, Gemini, Claude or local Ollama and LM Studio models.
Good for
- Prototyping multi-agent workflows before committing to a production design
- Running a local model for agents that should not send data to a hosted API
- Comparing how different LLMs perform within the same agent pipeline
- Building reusable agent workflows defined in configuration files
Pricing
Open source
Free and open source under GPL-3.0.
AgentForge FAQ
Which LLM providers does AgentForge support?
The README names OpenAI, Google Gemini and Anthropic Claude, plus local models via Ollama and LM Studio.
What license is AgentForge released under?
It is licensed under the GNU General Public License v3.0.
How are workflows defined?
Workflows are composed as Cogs using simple YAML files, according to the README.
Is AgentForge free?
Yes. AgentForge is open source, so you can use and self-host it for free.
What are the best alternatives to AgentForge?
The closest alternatives to AgentForge are Web, agent-framework, crewAI, CowAgent and Web.
Where this comes from
- Description: written from the project's README.


