
CloudBees has opened up a preview of an MCP Server for its nascent Unify platform to allow it to manage the AI agents that are working their way into DevOps workflows.
CloudBees’ Unify was only launched last month, as a platform for managing multiple DevOps platforms and tool chains, such as CloudBees-backed Jenkins and other CI/CD environments, as well as Git-based platforms such as GitHub and GitLab.
MCP, open sourced by Anthropic last year, has rapidly become the standard for connecting AI agents and broader systems.
Or as, CloudBees puts it, “to facilitate context-aware AI use in real-world enterprise systems.”
In practice, CloudBees said, its MCP Server will act as “A lightweight, standardized interface between large language models (LLMs)… and enterprise DevOps tools and processes.”
That means it will support functions such as pulling contextual data, such as pipeline, test and security data from Unify, as well as triggering CI/CD actions.
It should also allow users to “experiment with Unify … without needing to learn how to navigate the Unify APIs.”
CloudBees suggested it would allow agents to access enriched metadata, including DORA metrics or security posture as well as more conventional logs.
Together with Unify, it will allow workflow automation. Critically, it means that “actions” through the server are covered by enterprise level governance.
Just how far agents are being used in AI workflows may be open to debate. But developers are certainly widely adopting AI tools to generate code. GitHub last year said that 88 percent of US developers reported “at least some company support “ for such tools, while Cloudsmith said that of those that do use the tools, 42 percent said they accounted for half their code.
But Cloudsmith also reported that two-thirds of AI generated code was manually reviewed – a situation that may not be sustainable. Opening the way for AI agents to start checking the output of AI code generators.
This week Anthropic added support for remote MCP servers in its own code gen tool, Claude Code. This gives it the ability to “pull context from your third-party services — such as dev tools, project management systems, and knowledge bases—and take actions within those services.”
The announcement cites the example of integrating with the Sentry MCP server, to access errors and issues, “Then, you can debug using the context of those issues without leaving your terminal.”
Also this week, workflows and collab software purveyor Asana patched one of MCP’s first bugs – having first rolled out its MCP server on May 1. The org discovered a vuln in the MCP server on June 4 and took the feature offline from June 5 to June 17. Although Asana’s incident report didn’t provide details about the coding snafu, according to a notes to customers, the “bug could have potentially exposed certain information from your Asana domain to other Asana MCP users.”