ProDisco gives AI agents Kubernetes access that closely follows Anthropic’s Progressive Disclosure pattern [1]: the MCP server exposes search tools which in turn surface TypeScript modules, agents discover them to write code, and only the final console output returns to the agent.
ProDisco goes a step further: instead of exposing custom TypeScript modules, it provides a structured parameter search tool that returns the most suitable methods from the official Kubernetes client library, including the type definitions for their input and return values. This lets agents dynamically interact with the upstream Kubernetes library while avoiding any ongoing maintenance burden in this repository to mirror or wrap those APIs.
ProDisco goes a step further: instead of exposing custom TypeScript modules, it provides a structured parameter search tool that returns the most suitable methods from the official Kubernetes client library, including the type definitions for their input and return values. This lets agents dynamically interact with the upstream Kubernetes library while avoiding any ongoing maintenance burden in this repository to mirror or wrap those APIs.
https://github.com/harche/ProDisco
[1] https://www.anthropic.com/engineering/code-execution-with-mc...