Cortex is an open-source toolchain that converts API specifications and Markdown into interactive documentation, typed SDKs across eleven languages, and Model Context Protocol servers for AI agents.
What the product does and how it is positioned
Cortex serves as an API interface generator designed to bridge human developers and automated agents. Governed by a single configuration file, the system ingests multiple contract formats including OpenAPI, AsyncAPI, GraphQL, gRPC, and OpenRPC, verifying schema validity before generating project assets.
Through a unified command-line workflow, Cortex outputs an interactive documentation site with live testing capabilities, native SDKs across eleven distinct programming languages, and an MCP server that packages API endpoints as typed tools for agentic integration.
Source-supported ways to use the product
Engineering teams managing REST, realtime, GraphQL, or RPC interfaces can generate a unified, self-hosted documentation site with searchable references and interactive request tools.
Developers can generate Model Context Protocol servers to provide AI agents with typed tool schemas, embedded specifications, and client instructions directly from their API contracts.
API providers can compile client libraries in eleven programming languages featuring built-in retries, authentication, and streaming.
The documented workflow, where available
Specify input sources such as OpenAPI, AsyncAPI, GraphQL, gRPC, or OpenRPC files alongside MCP package settings in a configuration file.
Run the generation command to validate source contracts and compile interactive documentation, client SDKs, and the MCP server.
Deploy the interactive documentation site to a Node-compatible host, publish client libraries to language package registries, and connect the MCP server to AI agents.
Cortex translates API contracts into typed Model Context Protocol servers tailored for autonomous development agents. Each operation defined in the source specifications maps directly to a typed tool schema, eliminating ambiguity during automated tool selection.
Alongside tool definitions, the server bundles source specifications, documentation guides, and client library references. This structure gives connected agents grounded context to identify relevant endpoints, install appropriate SDK packages, and generate typed integration code.
Cortex unifies multiple API architectural styles into a single pipeline. Teams can combine REST, WebSocket, GraphQL, gRPC, and JSON-RPC definitions within one configuration, which Cortex validates prior to asset compilation.
The engine generates idiomatic client libraries in eleven languages, including TypeScript, Python, Go, Java, and Rust, incorporating standard network handling such as timeouts, retries, authentication, and streaming. Developers can also export and override Eta templates to alter generated outputs.
Checks to run with your own material and workflow
What was checked and when
Answers based on the source-checked product record
Cortex accepts OpenAPI, AsyncAPI, GraphQL, gRPC, and OpenRPC specifications, as well as Markdown documentation files, configured in a central project configuration file.
Cortex generates typed client libraries in eleven languages: TypeScript, Python, Go, Java, Kotlin, Ruby, PHP, C#, Rust, C++, and C.
Cortex generates a Model Context Protocol server that exposes API operations as typed tools, embedding API contracts, documentation, and SDK references to provide context for AI agents.
Yes, developers can export and override any of the underlying Eta templates to modify the structure and implementation of generated SDKs and documentation.
The generated interactive documentation site can be self-hosted and deployed to any hosting environment where Node runs.
