Warning: Guided is an experimental tool and is a work in progress. APIs, configuration, and behavior may change without notice. Use with caution.
Guided is a CLI tool designed to amplify the work of engineers using AI.
The idea is simple. Avoid spaghetti code and create maintainable code by making a solid scaffold.
- Provide a framework to execute in parallel
- Leverage local LLMs with instruction following and thinking
- Use tools in a containerized environment with Kubernetes
- Log everything and make it reversible
We believe that while work can be automated, engineers are an important piece of the puzzle when it comes to managing software.
While automation and artificial intelligence can be leveraged to write code quickly, managing the direction of a long term project will require a deeper level of understanding.
Properly planned code can go a long way to not only solve problems scaling but also increase velocity and maintainability.
The goal of Guided is to provide a framework for engineers to manage the direction of a project while leveraging AI to execute the work.
We recommend using Ollama; install it if you have not already. Install a model which supports tool calling and thinking.
Use uv to manage the Python virtual environment.
Install the CLI tool.
Run the installation script (e.g. - always check an installation script by downloading and reading it first)
curl "https://raw.githubusercontent.com/guided-code/guided/refs/heads/main/bin/install" | /bin/shConfigure your environment by running:
guide configureStart a TUI and chat directly
guide chatOr use a command line interface
echo "What is 23 * 17?" | guide chatCurrently setup to use Ollama to run local agents.
Find documentation in the docs folder.
We encourage you to contribute to Guided. Guided is about building great tools to enable engineers to build more. Join us! Read about contributing
Guided is licensed under the MIT License.
