-
Notifications
You must be signed in to change notification settings - Fork 208
Description
Abstract: I propose adding a new module hawkbit-mcp-server to the project. This module implements the Model Context Protocol (MCP), allowing LLMs (like Claude, OpenAI, Gemini) and AI agents to natively interact with the Eclipse Hawkbit Management API.
Motivation: As AI agents become more prevalent in DevOps and IoT operations, there is a need for a standardized way to expose Hawkbit's capabilities to these models. The MCP standard provides this interface. This enables use cases such as:
- "Update all targets in the 'Warehouse' tag with the latest firmware."
- "Show me why the rollout Bug/target table perfomance #105 is stuck."
- "Create a new distribution set and assign it to compatible target types."
- Extract more strategic or specific information about the targets or about the rollouts/actions.
Proposed Architecture: The module is built using Spring AI (spring-ai-starter-mcp-server) and leverages the existing hawkbit-sdk-mgmt.
I'm implementing a Proof of Concept (POC) in a personal public repository, and the initial tests were very promising, but of course, it still needs refinement to increase efficiency by consuming fewer tokens, as well as standardization, documentation, and security.
https://github.com/Romulohenriquesc/hawkbit-mcp-server
If you prefer, I could create a preliminary demo PR, but my repository is public for anyone to test.