Build and Manage AI Agents from Your Terminal

If you’ve been experimenting with AI agents, you know the drill: switch between a browser, a notebook, and your terminal, copy-pasting commands and managing scripts. It’s functional, but it breaks your flow. What if you could stay in your terminal—the developer’s home turf—to build and control these agents directly?

That’s exactly what Incur offers. It’s a new CLI tool that lets you orchestrate AI-powered agents and workflows without leaving your command line. For developers who live in the terminal, this is a promising shift towards a more integrated and scriptable AI development experience.

What It Does

Incur is a command-line interface for creating and managing AI agents. Instead of relying on a separate dashboard or web app, you define agents, their tasks, and their workflows using straightforward CLI commands or configuration files. You can run agents, chain tasks, and monitor outputs, all within your terminal session. It’s essentially bringing the building blocks of AI automation into the shell environment where many developers are already most productive.

Why It’s Cool

The core appeal is the integration into the terminal-native workflow. This isn’t just a thin wrapper around an API; it’s built for developers who think in commands and pipelines.

• Terminal-First Design: Everything is done via the CLI. You can quickly spin up an agent, run a one-off task, or compose multi-step workflows using commands that feel right at home next to git or npm.
• Scriptable and Automatable: Because it’s in the terminal, Incur agents can be easily integrated into shell scripts, CI/CD pipelines, or cron jobs. You can pipe outputs to other Unix tools, log results to files, or trigger agents based on other system events.
• Developer-Centric Configuration: You can likely define agents in a config file (like incur.json or incur.yml), making it easy to version control your agent setups and share them with a team.
• Reduced Context Switching: Keeping the agent lifecycle inside your terminal means less alt-tabbing and a more focused development loop. You can edit code, run an agent, and see the results in the same window.

How to Try It

Ready to take it for a spin? The project is open source on GitHub.

1. Head over to the repository: github.com/wevm/incur
2. Check the README for the latest installation instructions. It’s likely as simple as running an npm install or using a package manager like brew.
3. Follow the quick start guide to define your first agent and run a task.

The docs will guide you through setting any required API keys and your initial configuration.

Final Thoughts

Incur taps into a real need for developer-friendly AI tooling that doesn’t force a new ecosystem. By placing agent control in the terminal, it empowers developers to build automation into their existing workflows naturally. It’s early days, but the concept is solid. If you’re already tinkering with AI agents and wish you could manage them with the same simplicity as a background process, Incur is definitely worth a look. It might just become a staple in your toolkit for prototyping and deploying lightweight AI automations.

Follow us for more cool projects: @githubprojects