What it does
Octri imports an OpenAPI, Swagger, Postman or AsyncAPI spec in JSON or YAML and normalises it to OpenAPI 3.1. Users connect a spec by uploading a file or linking a GitHub repository. AI drafts documentation for every endpoint. The user edits the draft and publishes the documentation to their own domain. Octri generates client SDKs in ten languages from the same spec. It also hosts a remote MCP server at one URL. Monitoring covers errors, logs, traces and uptime checks. Every spec change reruns the whole pipeline for documentation, SDKs and agent tools.
Capabilities
- The SDKs cover TypeScript, Python, Go, Java, Dart, Ruby, PHP, Rust, Swift and Kotlin, with retries, pagination and typed errors built into every client.
- SDKs can auto-publish to linked registries, including npm, PyPI, Maven, crates.io and RubyGems, when registry auto-publish is on.
- The MCP server offers a tool for every endpoint plus seven tools that search and read the documentation. OAuth sign-in lets each agent call the API as the person who connected it.
- Monitoring groups errors by fingerprint and sends alerts to Slack or any webhook. Monitoring is off by default, and each SDK user must opt in.
- The documentation includes a live try-it playground that calls the real API with the reader's own key.
When to use it
- You want documentation, client libraries, an agent-facing MCP server and monitoring generated from one OpenAPI spec.
- You want error tracking inside the SDKs you ship and monitoring for your API.
- You are moving from another documentation or SDK tool and want your existing configuration read and brought across.
