A new open-source tool called Henka has emerged on GitHub, positioning itself as a multi-tenant MCP (Model Context Protocol) server purpose-built for structured, semantics-aware code refactorings. The project, hosted at github.com/martint/henka, appeared on Hacker News where it attracted modest attention with just 4 points and no visible comments at time of publication. The core innovation here is the multi-tenant architecture applied to code transformation tasks. Rather than each AI assistant or IDE managing its own refactoring logic in isolation, Henka provides a centralized server that multiple clients can query for semantic-aware code changes. This approach could simplify tooling stacks for teams running multiple AI coding assistants or development environments. The "semantics-aware" descriptor suggests the tool goes beyond simple text manipulation or regex-based transformations. By leveraging MCP, the protocol designed specifically for connecting AI models to external tools and data sources, Henka can potentially access deeper code understanding from language servers or static analysis engines before proposing refactorings.

Infrastructure Considerations

For infrastructure teams evaluating this project, the multi-tenant design raises interesting questions about isolation, resource allocation, and security when multiple users or agents share refactoring capabilities. The server model means one place to update transformation rules, but also introduces a potential bottleneck and dependency point that teams would need to monitor.

Key Takeaways

  • Henka implements MCP protocol for standardized AI-tool communication around code changes
  • Multi-tenant architecture allows shared infrastructure for refactoring across multiple clients
  • Semantics-aware approach aims for transformations that understand code structure rather than just text patterns
  • Project is early-stage with limited community feedback visible on Hacker News

The Bottom Line

Henka addresses a real pain pointβ€”coordinating intelligent code changes across AI-assisted development toolsβ€”but the multi-tenant MCP refactoring space is nascent. Teams interested in this approach should evaluate whether centralized transformation logic fits their workflow before committing to the architectural pattern.