Uber’s MCP Gateway Connects AI Agents to More Than 5,000 Internal Tools

As reported on November 18, 2026, Uber had built an MCP Gateway to standardize how AI agents discover and use internal services. The platform supports more than 800 MCP servers and 5,000 tools, addressing fragmented integrations, duplicated infrastructure and inconsistent security practices.

The architecture separates a MCP Registry control plane from a Proxy Gateway that handles requests. The registry stores tool definitions, schemas, ownership and enablement settings. The proxy translates MCP requests into HTTP, gRPC or TChannel calls, then converts the responses into MCP-compatible results. Uber engineers describe this as a way to expose existing APIs as MCP tools without changing the underlying services.

Uber’s AutoCrawler uses Cadence to scan internal interface definition language registries. It extracts API definitions from Protocol Buffers and Thrift schemas, generates MCP-compatible schemas and uses a large language model to write tool descriptions. Newly discovered tools remain disabled until their owning teams review and explicitly enable them. Configuration changes can reach the data plane without a service restart or redeployment.

Uber’s MCP Gateway Connects AI Agents to More Than 5,000 Internal Tools

The gateway uses Uber’s internal access-control system to authorize access at both the server and tool levels. It also supports redacting personally identifiable information and other sensitive data from tool responses. For third-party integrations such as Jira and Google, a separate service exchanges internal user tokens for the appropriate third-party authentication tokens.

To avoid loading thousands of tool definitions into an agent’s context, Uber developed Omni MCP for progressive discovery: agents can search for servers, find relevant tools, retrieve an individual schema and invoke a selected tool when needed. Response Projection lets agents request specific response fields rather than a full payload. Uber’s aifx command-line interface also lets coding agents discover and invoke tools, writing results to files for selective inspection.

In LinkedIn comments cited in the report, Microsoft CoreAI principal software engineer Ajay Sharma and Pipe17 president Kelly Goetsch compared the approach with established API gateway patterns. Tamir Dresher likewise pointed to the continuing importance of security, service discovery, observability and governance in distributed systems.