Native MCP support in BluBees: what it means for AI Agents in Jira

[#1]

We just launched our Beta release for BluBees, and the headline capability is native MCP Tool Support for our AI Agents. I thought it was worth sharing what we built and how it works in practice, specifically within the Atlassian ecosystem.

What MCP Tool Support means for AI Agents in Jira

Until this release, a BluBees AI Agent could only use tools from our pre-built Connector library, a useful set, but with a hard boundary. If a tool wasn't in the library, the Agent couldn't reach it.

With native MCP support, that boundary is gone. Any application that exposes an MCP server is now available as a tool skill for any BluBees Agent, without custom integration work. The Agent discovers the tool's capabilities through the MCP server and invokes them as part of any Flow, in real time, with no middleware.

You can configure multiple MCP servers simultaneously, which means a single Agent can have tool skills across several MCP-connected applications at once. And because the connection is standardised through MCP rather than bespoke per tool, adding a new application to an Agent's reach is a configuration step, not a development project.

The Atlassian and Forge angle

Because BluBees runs natively on Atlassian Forge, all MCP communication happens within the Atlassian cloud boundary, controllable at the MCP server level. There's no data leaving the Atlassian environment to reach an external MCP server. For enterprise and government teams where data sovereignty is non-negotiable, that's a meaningful distinction compared to MCP implementations that sit outside the Atlassian ecosystem.

What else is in the Beta

Alongside MCP, the Beta release includes:

• Anthropic Claude and DeepSeek added natively: four AI models now available, choose per Agent per Flow
• Webhook Node: connect to any external system from any point in a Flow, with response capture and downstream passing
• Visual Conditional Route Builder: runtime data evaluation for Flows that handle real-world, unpredictable data
• TestRay Secured Gateway: API token generation and direct connection within BluBees
• Canvas overhaul: Live Log download, Draft/Active status, Tile View, right-click context menu

[External links removed]

We're currently in Beta with select teams download the app from the Atlassian Marketplace.

For teams exploring MCP in the Atlassian ecosystem, what use cases are you finding most compelling? And are there tools you've been unable to connect to your AI workflows that MCP would unlock?