Octri SDK Generator：从 OpenAPI 规范生成文档、SDK、MCP 服务器与监控

Your OpenAPI spec, all the way to production.

A docs site, SDKs in ten languages, an MCP server for agents, and monitoring that tells you when those SDKs break in production. One spec generates all four and keeps them current.

(3 min video)

One input

Upload a spec once. Ship four products

Docs, SDKs, monitoring and agent tools. Each one stands on its own. They just happen to be better together.

1. input

   Your spec

   Any format: OpenAPI, Swagger, Postman or AsyncAPI, JSON or YAML. Normalised to OpenAPI 3.1.

2. Docs site

   Written, edited, published

3. SDKs · 10 languages

   Tuned to your house style

4. Observability

   What breaks, and why

5. MCP server

   Your API as agent tools

Four products, one platform

API StudioAPI documentation

Docs worth reading, from the spec you already have

AI writes the first draft for every endpoint. You edit it like a document. Nobody has to open the YAML.

• Three-column endpoint pages with schema trees that open a level at a time
• Live “try it” playground on every endpoint
• MDX guides alongside the generated reference
• Your own domain on every tier, Free included

See API Studio

SDK StudioSDK generator

Ten languages, built to your conventions

Rename methods, exclude endpoints, pick your HTTP engine and folder structure. Then let it publish itself.

• Per-language config for namespaces, pagination, idempotency and code style
• Custom hooks you write, compiled into the client
• Auto-publish to npm, PyPI, Maven, crates.io, RubyGems and more
• Build history so you can see exactly what shipped and when

See SDK Studio

SDKs are generated for TypeScript, Python, Go, Java, Dart, Ruby, PHP, Rust, Swift, Kotlin.

MonitoringSDK and API monitoring

Find out what breaks before your customers tell you

Two ways in: flip it on and the telemetry compiles into your generated SDKs, or drop the standalone package straight into your backend. Either way, there's no agent to deploy and nothing to instrument.

Octri Monitoring: querying structured logs, opening a raw event payload, grouped issues, a trace waterfall across client, server, cache and database, the service map, N+1 query detection, synthetic uptime checks and alert rules.

Monitoring is off by default and switches on from the dashboard. Credentials and direct identifiers, from tokens to email, phone and IP, are redacted on the client side, before an event leaves your process, then again at ingest. Personal context waits on your app's consent, under a pre-signed GDPR Article 28 DPA.

MCPMCP server for agents

Stop agents from hallucinating your API

MCP is how an agent reaches something it was never trained on. Without it, Claude or Cursor integrates your API from memory. With it, one command turns your project into a server they connect to. Then they list your real endpoints and call them.

• Seven documentation tools to search, read and navigate your docs
• One callable tool per endpoint, so the agent can hit your API for real
• Curated by SDK Studio, so your exclusion list becomes the agent's permission list
• One line to install: npx @octri/mcp, with nothing to host

See MCP

Is your spec ready for code generation?

Paste your OpenAPI URL and get a score out of 10, on the same rules the SDK Studio runs. You see every missing schema, undeclared path parameter and awkward method name, before anyone generates a client from it.

An example audit runs alongside: a readiness score of 6.8 out of 10 across 24 operations, with the rules the spec passed and failed scrolling past it.

What the fourteen rules check, and what each one costs a generated SDK.

Change one. All four follow

Deprecate an endpoint in API Studio and the SDKs mark the method, the agent tools stop offering it, and monitoring shows you who still calls it. Add a language, cut a release, push a new spec: the same thing happens. There is one source underneath all four, so there is never a second place to go and update.

Everyone else stops at “here's your SDK.”

That's the moment your code leaves your visibility. It runs on someone else's machine, fails on someone else's machine, and you hear about it in a support ticket three days later.

Without Octri

Error thrownon their machine

Nothing reportedno alert

3days

Still broken in prodyou have no idea

Angry support tickets#4412-4414

and again

With Octri

Error thrownon their machine

Alert firedinstantly

9min

Fix shippedv2.3.2

0 support ticketsnobody noticed

From spec to shipped in an afternoon

The whole pipeline from one OpenAPI spec: API documentation, client libraries in ten languages, an MCP server and production monitoring.

01

Connect your spec

Upload a file or link a GitHub repo. Coming from Postman or AsyncAPI? We translate it.

02

Review the draft

AI writes docs for every endpoint. Edit anything, publish when it reads right.

03

Pick your languages

Configure the SDKs, flip on monitoring, build. They publish themselves.

04

Push to repository

Every spec change reruns the whole pipeline. Docs, SDKs, agent tools, all in sync.

One spec, ten client libraries

Every client is generated from the same OpenAPI document and published to the registry its own users already install from.

• TypeScript

  Publishes to npm with the type definitions generated from your spec. fetch or axios.

• Python

  Publishes to PyPI. Async first, with optional sync variants and a choice of data model.

• Go

  Distributed by git tag, so go get resolves it straight from your repository. Standard library HTTP.

• Java

  Publishes to Maven Central, signed, under a groupId on a domain you own.

• Dart

  Publishes to pub.dev, for Dart and Flutter alike.

• Ruby

  Publishes to RubyGems, with the class-style client Ruby callers expect.

• PHP

  Publishes to Packagist, so Composer installs it like every other dependency.

• Rust

  Publishes to crates.io, and full doc comments become the rustdoc your users read on docs.rs.

• Swift

  Distributed by git tag and resolved by Swift Package Manager. No registry account.

• Kotlin

  Publishes to Maven Central. OkHttp or Ktor, whichever your team already runs.

Already on something else?

We read your existing config and bring it across: navigation, custom pages, SDK settings and all. You don't start from a blank project. Want a hand with the move? A call with us and our onboarding help are both free of charge.

Simple, combined pricing

Docs, SDKs and monitoring in one plan. Extra SDK languages are a flat add-on.

Extra languages are a flat $50/mo.

Recommended

Growth

$79/mo+ tax

For growing teams shipping fast.

Choose Growth

Enterprise

Custom

Unlimited scale with SLA guarantees.

Contact sales

Notes from building it

What we learn shipping specs, SDKs and telemetry, written up properly.

All articles

Questions

Plans, spec changes, editing the generated docs and SDKs, training data and self-hosting.

No. Plans are not per-product, so you can leave one switched off and turn it on months later without redoing anything you have already set up.

Your spec is already written

Everything else takes an afternoon.