Plug your coding agents into Tuist
With coding agents increasing the speed and volume of code being produced, it has never been more important to ensure a healthy project setup where agents and developers can validate their work efficiently. There is nothing more frustrating than a feature or a fix taking longer to reach users because of a slow compilation, a flaky test run, or a build artifact that quietly regressed.
At Tuist, we realized we already had most of the ingredients to fix this.
- We built, maintain, and continuously scale a decentralized global cache infrastructure that gets as close as possible to the offices and environments where the work happens.
- We observe builds, tests, and built artifacts, and track how they evolve over time and across environments.
The only pieces we were missing were the ability to act on that data intelligently. One we did not need to build ourselves: LLMs already have the knowledge to reason over it. The other we did need to build: making the data available to them. That is what we have spent the past weeks working on.
Agents as project doctors
When we ran experiments giving agents access to Tuist’s data, the results were immediate. Agents can reason over your build and test history, spot regressions, identify bottlenecks, and help you maintain a productive development environment. Think of them as doctors for your project, with Tuist as the diagnostic layer underneath.
To make this possible, we have exposed all of that data through:
- REST API
- MCP server
- CLI commands
- A set of skills to guide agents through the most common scenarios
Whether you are using Codex, Claude, opencode, or any other coding agent, you can plug it into Tuist via any of these interfaces and start getting value immediately.
See it in action
We did the scripting and structure for this video in the open, as we like to do: Marketing video: Introducing MCP. We plan to keep doing it that way.
You can also read the full blog post here: tuist.dev/blog/2026/03/18/doctor
What’s next
We will keep scaling the infrastructure and the data we collect, which directly expands what agents can do with your projects. As always, keep sharing your use cases with us and we will figure out how to support them.
But we are not planning to stop at reactive insights. Once we have validated this foundation, we want Tuist to take a more active role in keeping your projects healthy over time, not just answering questions but flagging problems before they slow you down. More on that soon.