Mngr is a command-line tool designed to manage and scale AI coding agents, enabling users to run hundreds in parallel for automated workflows like fixing tests and opening pull requests. It supports various agents and compute platforms, providing real-time monitoring and interaction capabilities. As a free and open-source tool, it facilitates efficient development processes without vendor lock-in.
Mngr
Run 100s of Claude agents in parallel
Mngr 介绍
核心功能
- Run hundreds of AI agents simultaneously in parallel
- Compose and automate complex workflows with programmatic control
- Real-time agent monitoring, including status checks and mid-task debugging
- Fast agent startup under 2 seconds and idle shutdown for resource efficiency
- Platform-agnostic operation across local, Docker, Modal, or SSH environments
典型使用场景
- Software developers automating code testing, bug fixes, and code reviews
- DevOps engineers scaling validation and deployment pipelines with multiple agents
- AI researchers conducting parallel experiments and simulations using agent instances
- Open-source maintainers automating issue triaging and pull request creation
- Teams building AI-driven development and quality assurance workflows
为什么适合初创团队
Startups need Mngr to automate development tasks and scale operations efficiently with limited resources, accelerating processes like code testing and deployment through parallel agent execution. Its free and open-source nature reduces costs and allows for customization to fit specific workflow needs, enabling faster iteration and innovation.
可替代选择
LangChain, AutoGPT, Hugging Face Transformers, Docker Compose
常见问题
Is Mngr free to use?
Yes, Mngr is completely free and open-source under the MIT license.
What types of AI agents does Mngr support?
It primarily supports Claude Code but is designed to work with any agent harness, including other AI models and custom implementations.
Can Mngr run on cloud infrastructure?
Yes, it can run on various platforms such as localhost, Docker, Modal, or any SSH-accessible host, making it flexible for cloud deployments.
How does Mngr handle agent isolation and security?
It provides proper isolation of settings and state files to ensure multiple agent instances run independently without interference.
