Vibe Boldly. Verify Everything.

vaibify creates secure, containerized environments for AI-assisted data analysis that can be accessed through a web browser. It fully embraces agentic AI code development, but recognizes that a human must verify all results. vaibify uses Docker containers to prevent AI agents from harming your sensitive data. These containers can be monitored and modified through an application that includes terminal window(s) for running agents like Claude Code, codex, or Gemini and "viewing windows" for inspecting results (data files, figures, animations). Work with agents to be creative in a sandbox, develop a toolkit, or enter "Project" mode, which enables pipeline development with automated and manual verification tracking for each step. vaibify is vigilant, alerting you to changes in your dependencies, so when your agent edits a critical file that updates an critical file, you immediately know all the downstream consequences. Seamlessly link your work with external resources like GitHub, Overleaf, and Zenodo for monitoring software development, writing reports, and archiving your results. vaibify allows you to vibe code with confidence: your host machine stays safe while the agents freely develop code and build your analysis pipeline — all with minimal IDE interaction — enabling you to focus on vetting the results via visual inspection, writing up a summary, and acting on the new insight. vaibify takes the pain out of making science reproducible: it preserves the environment and the procedure, gets as close to byte identity as it can, and makes every remaining difference inspectable.

In this screenshot of the vaibify dashboard, the steps to your workflow are tracked on the left. View the contents of the vaibify container along the top row in "viewing windows". Manage your agents and navigate the container yourself in terminal window(s) in the bottom of the GUI. Use buttons and menus to perform most basic tasks, or ask your agent to make changes. Additional pages allow you to create and manage containers and projects (see documentation).

Installing vaibify takes a few minutes. Containers also need Docker (Colima on macOS), and the first time you open a container project vaibify builds its image, which can take from several minutes to over an hour. See the full documentation for installation instructions, CLI reference, configuration, security model, and contributor guidelines. But you can get started without the container if you just want to quickly try it out. Read the Quick Start Guide, then just run vaibify to launch the GUI that will guide you through building containers, creating workflows, syncing with external services, and verifying your vibe-coded scientific workflows.

Found a bug or something confusing? Please open an issue. Two things make a report much easier to act on: the output of vaibify doctor (a pre-flight check of your Docker environment), and the relevant lines from the host log at ~/.vaibify/vaibify.log.

If you use vaibify in your research, please consider citing Barnes, R. (2026) "Integrating AI Agents into Scientific Research."

© 2026 Rory Barnes.