Octri is a platform for the entire API production workflow, with the core idea of "one OpenAPI specification, four products generated." Users point their API specification (OpenAPI, Swagger, Postman, or AsyncAPI, in JSON or YAML) at Octri, and the platform normalizes it to OpenAPI 3.1 and automatically generates and maintains the following four types of artifacts: a documentation site, SDKs in 10 languages, an MCP server, and production monitoring. Each product can be used independently, but they work better when combined.
Key Capabilities
Documentation site: AI generates a first draft for each endpoint, users can edit it just like editing documentation, and publish it to their own domain. Endpoint pages use a three-column layout, include a progressively expandable schema tree, a live "try it" playground for each endpoint, and support adding MDX guides alongside the generated reference documentation. All plans (including the free version) support custom domains.
SDK generation (10 languages): Supports TypeScript (published to npm, using fetch or axios, with type definitions generated from the specification), Python (published to PyPI, async-first, with optional sync variant and data models), Go (distributed via git tags, using the standard library HTTP), Java (published to Maven Central, signed, using a groupId with its own domain), Dart (published to pub.dev, suitable for Dart and Flutter), Ruby (published to RubyGems, class-style client), PHP (published to Packagist, installed via Composer), Rust (published to crates.io, with doc comments generating rustdoc), Swift (distributed via git tags and Swift Package Manager, no registry account required), Kotlin (published to Maven Central, with optional OkHttp or Ktor). Supports configuring namespaces, pagination, idempotency, and code style per language, custom hooks can be compiled into the client, automatic publishing to registries such as npm, PyPI, Maven, crates.io, and RubyGems, and build history is provided.
Monitoring and observability: Users can choose to enable Monitoring in their project and rebuild, allowing SDK users to voluntarily enable it from the application; or add a dedicatedInstant setup. No independent agent needed. Query structured events by level, route, status code, or release version; group errors by fingerprint and link them to the function and file that threw the error; trace a single request end-to-end across client, server, cache, database, and queue, and flag slow spans; display service-to-service calls and the error rate for each edge; automatically discover identical queries in a loop and count them in traces; run synthetic probes on a schedule and retain historical records; trigger alerts based on burn rate and regressions, avoiding a single 500 error waking everyone. Credentials and identifiers are redacted before client events leave the process and at ingestion. Monitoring is off by default, enabled from the dashboard, personal context requires app consent, and it follows the pre-signed GDPR Article 28 DPA.
MCP server: a single command, npx @octri/mcp, turns a project into an MCP server with no hosting required. It provides 7 documentation tools for searching, reading, and navigating docs, with one callable tool per endpoint, letting MCP clients such as Claude and Cursor list and call real endpoints. SDK Studio's exclusion list becomes the agent's permission list.
Spec readiness score: paste an OpenAPI URL to get a score out of 10, based on the same rules SDK Studio runs, showing missing schemas, undeclared path parameters, and non-standard method names. A sample audit shows 24 operations scoring 6.8 and lists the 14 rules that passed and failed and their impact on generated SDKs.
How to use
- Connect a spec: upload a file or link a GitHub repository; Postman or AsyncAPI formats are translated.
- Review the draft: AI writes documentation for each endpoint, users can edit it, and publish when satisfied.
- Choose languages: configure and build SDKs. Monitoring requires each SDK user to opt in voluntarily; a linked registry can publish automatically.
- Push to the repository: every spec change re-runs the entire pipeline, keeping docs, SDKs, and agent tools in sync.
The platform supports migration from existing configurations, reading existing navigation, custom pages, SDK settings, and more, so you don't have to start from a blank project; migration assistance and onboarding help are free. Self-hosted generation and documentation rendering are not yet available; this is currently a hosted service.
Use cases
Suitable for teams that need to maintain API documentation, multi-language SDKs, and monitoring, especially developers who want to expose API endpoints to AI agents such as Claude and Cursor. Fromspec to delivery, a normal documentation site takes about a day, and the rest can be done in an afternoon.
Limitations
This is currently a hosted service; self-hosted generation and documentation rendering are not yet available. Custom SDK code is only available on the Growth plan and above. Personal context requires app consent and follows the GDPR Article 28 DPA. SDK monitoring requires SDK users to opt in voluntarily, or you can add an independent package on your own backend. Specific plan tiers, pricing details, and the feature scope of the free version are not yet provided.
