amoylab/unla is a gateway that turns already-running MCP servers and RESTful APIs into services that follow the same specification (MCP Protocol). The conversion happens through YAML configuration alone, and the repository description states that neither the target code nor the infrastructure has to be changed. In the available record, the most recent release tag is v0.10.0, and the repository was created on April 15, 2025. However, the given record contains no changelog for v0.10.0, so what exactly changed in this version cannot be determined.
The core is protocol translation. It converts RESTful APIs into MCP servers without code changes, and proxies MCP services that are already running. It supports the MCP SSE and Streamable HTTP transports. It also supports multi-tenancy, where several teams share one gateway, persistent and recoverable sessions that bring broken sessions back, and an admin UI. Deployment supports Docker, and Kubernetes and Helm deployment are supported as well. The license is MIT, the primary language shown by GitHub is TypeScript, and the README states that the service is implemented in Go. The repository has 2244 stars.
What needs checking is the strength of the evidence. The available record contains no figures for the latency and throughput it achieves, and no comparison results against other gateways. The momentum score in the research record is 0.8 out of 10, and the criteria behind it as well as the novelty, technology, and quality scores are not given. If your team has to add MCP without touching existing infrastructure, this is worth a look, but judging on the basis of performance and scale is difficult from the published facts alone.