We've all been there. You build an API, you're proud of it, and then someone asks:
"Hey, where are the docs?"
And you paste them a GitHub link to a markdown file.
It works — technically. But it's not great. There's no try-it button, no version history, no search, no way for an AI assistant to understand your endpoints. Just a wall of text.
There's a better way, and it takes about 5 minutes to set up.
What APIKumo publishes
APIKumo is a full API workspace — you design requests, automate auth, version your work, and publish typed docs to your own subdomain. The publish step is what this post is about.
When you hit Publish on a collection, APIKumo gives you:
- 🌐 A live subdomain —
your-api.apikumo.com(or your own domain) - 🔍 Full-text search across every endpoint
- 📜 Scroll-spy navigation so readers always know where they are
- 🎨 Custom theme — your brand colors, logo, and name
- 🔢 Version selector — v1.0, v1.1, v2.0 all live at once
- 📋 Changelog so consumers know what changed between versions
- ▶️ Try-it — readers can fire real requests right from the docs (with their own auth)
- 🤖 AI docs chat — a built-in assistant that answers questions grounded in your endpoints
- 🔌 MCP endpoint — Claude, Cursor, and Continue can read and call your API automatically
How it works in practice
1. Build your collection normally
In APIKumo, you send requests the usual way — method, URL, headers, body, auth. The Monaco-powered editor handles JSON, XML, GraphQL, and more.
What makes it different: you also define request and response schemas — fields, types, enums, and examples. These become the docs.
GET /users/:id
→ 200 { id: string, email: string, role: "admin" | "member" }
→ 404 { error: string }
2. Snapshot a version
When your API is ready to ship (or already shipped), you take a version snapshot:
v1.0 → snapshot → "Initial release"
You can keep editing in v1.1 without touching the published v1.0. Consumers on v1 see stable docs; v2 readers see the new surface. No separate repos, no copy-pasting.
3. Write your changelog
Each version gets a changelog entry. Plain text or markdown — whatever your team prefers.
## v1.1
- Added pagination to GET /users
- Deprecated X-Auth-Token header (use Authorization: Bearer instead)
- Fixed schema for POST /events body
Your API consumers see this in the version selector. They know exactly what changed and when.
4. Hit publish
One button. APIKumo generates the full docs site from everything you've built — schemas, examples, descriptions, changelogs — and deploys it to your subdomain.
The result looks like professional docs. Because it is.
The AI layer on top
Every published collection automatically gets two things that feel like magic the first time you see them:
Docs chat: An AI assistant trained on your actual endpoints. Not generic "how do I use an API" answers — specific answers like "What fields does POST /webhooks accept?" or "Show me how to paginate through orders."
MCP endpoint: A Model Context Protocol URL your team can plug into Claude, Cursor, or Continue. The AI reads your API surface and can help write the integration code — without hallucinating endpoints that don't exist.
Why this matters more than a README
| README | APIKumo docs |
|---|---|
| Static text | Live, always up to date |
| No versions | v1.0, v1.1, v2.0 in one place |
| No try-it | Readers can fire requests in-browser |
| No search | Full-text search |
| No AI | AI chat grounded in your endpoints |
| No MCP | Claude / Cursor can call your API |
If you're building APIs and still shipping documentation as a markdown file, give APIKumo's publish flow a try. It's free while in preview.
🌐 apikumo.com · Sign in with Google, GitHub, or Discord.
Top comments (0)