From where you areto where AI meets you.

Three starting points. Three paths. Each one starts from what you already have and ends with AI using it — in under a minute.

Start with Google Start with GitHub

Try for Free·No credit card

ENTRY POINTS

CHOOSE YOUR PATH

Any system. Any expertise. One click.

Whether you connect an existing API, build with TypeScript, or write what you know — every path ends in the same place: a production MCP server, secured, deployed, and ready for any AI assistant.

Entry Points3 paths → 1 output

APIConnect · Zero code

CodeBuild · TypeScript

SkillsWrite · Publish

MCP Server → V8 Sandbox → Global Edge → Any AI Client

DEPLOY ONCE. USE INSTANTLY WITH YOUR FAVORITE AI CLIENTS:

ClaudeDesktop & Code

CursorIDE

ChatGPTOpenAI

WindsurfIDE

VS Code+ Copilot

ClineTerminal

PATH_01:API

You have a system with an API. Here's how AI starts talking to it.

THE API PATH

Paste. Configure.
Done.

STEP 01

Find your OpenAPI spec

Every modern system that has an API also has a spec document — a file that descr…

STEP 02

Paste the URL

In the Vinkius dashboard, choose 'OpenAPI' and paste your spec URL. Vinkius read…

STEP 03

Configure security

Add your API credentials — API key, Bearer token, or Basic auth. They're encrypt…

STEP 04

Deploy

One click. Your API is now an MCP server — live on the global edge. No infrastru…

"Your CRM is now an AI capability. Zero code. Zero infrastructure."

Paste → Configure → Deploy → Ask.

1. FIND YOUR OPENAPI

Find your OpenAPI spec

Every modern system that has an API also has a spec document — a file that describes what your API can do. It's usually at a URL like your-software.com/api/docs or your-software.com/swagger.json.

OpenAPI URL

https://your-crm.com/api/docs

✦ 47 endpoints found

Not sure where your spec is? Check your API documentation page, look for a 'Swagger UI' or 'OpenAPI' link, or ask your development team for the spec URL.

Salesforce, HubSpot, Stripe, Zendesk — any REST API with documentation.

2. PASTE THE URL

Paste the URL

In the Vinkius dashboard, choose 'OpenAPI' and paste your spec URL. Vinkius reads it automatically and detects every endpoint your system exposes.

OpenAPI URL

https://api.company.com/openapi.json

Server Name

Production ERP

Auto-filled from spec

Base URLOPTIONAL

https://api.company.com/v1

Auto-detected from spec

3. CONFIGURE SECURITY

Configure security

Add your API credentials — API key, Bearer token, or Basic auth. They're encrypted with AES-256 and never leave the vault.

✦ Security detected: Bearer Token

Auth Type

Bearer Token▼

Token

••••••••••••

⛨Encrypted with AES-256. Never leaves the vault.

4. DEPLOY

Deploy

One click. Your API is now an MCP server — live on the global edge. No infrastructure to manage, no containers to configure.

Parsing OpenAPI specs…

Sealing V8 Isolate Sandbox…

Injecting Zero-Trust DLP…

Distributing to Global Edge…

5. CONNECT YOUR AI

Connect your AI assistant

Copy the MCP URL. Paste it into Claude, Cursor, ChatGPT, or any compatible AI client. Ask a question — the AI queries your API directly.

Production ERP

https://cloud.vinkius.com/mcp/abc123

Raw

{ "email": "user@org.com" }

Redacted

{ "email": "u***@***.com" }

6. MONITOR AND GOVERN

Monitor and govern

Track every tool call, token cost, and data flow from the dashboard. DLP redaction, usage alerts, and FinOps controls — all in real-time.

Tool Calls

1,247

↑ 12% this week

Tokens

842K

within budget

⛨DLP active · 38 fields redacted today

PATH_02:CODE

You write TypeScript. Here's how your logic becomes a production MCP server.

THE CODE PATH

Create. Code.
Deploy.

STEP 01

Scaffold

One command creates a standard TypeScript project. No magic. No custom compiler.…

STEP 02

Write your tools

Tools are plain TypeScript functions with type-safe inputs. Define what the AI c…

STEP 03

Add middleware

Lifecycle hooks for auth, logging, rate limiting. Standard middleware pattern — …

STEP 04

Deploy

Live on the global edge in 4 seconds. Zero cold starts. V8 sandbox per request. …

"Your logic. Your types. Your tests. Production in 4 seconds."

npx vurb create → code → vurb deploy.

1. SCAFFOLD

Scaffold

One command creates a standard TypeScript project. No magic. No custom compiler. Just src/, package.json, and tsconfig.json.

$npx vurb create my-server

Project Structure

my-server/ ├── src/ │ ├── tools/ │ ├── presenter.ts │ └── index.ts ├── package.json └── tsconfig.json

2. WRITE YOUR TOOLS

Write your tools

Tools are plain TypeScript functions with type-safe inputs. Define what the AI can do. The Presenter layer controls what the AI sees — strip nulls, slice arrays, cap tokens.

.tstools/get-user.ts

1 server.tool('get-user',2 {'{'} id: z.string() {'}'},3 async ({'{'} id {'}'}) => {'{'}4 const user = await db.findUser(id)5 return presenter.format(user)6 {'}'})

3. ADD MIDDLEWARE

Add middleware

Lifecycle hooks for auth, logging, rate limiting. Standard middleware pattern — no framework surprise. Works with Express, Fastify, or standalone.

onRequest

→

onToolCall

→

onResponse

→

onError

onRequest → onToolCall → onResponse → onError

4. DEPLOY

Deploy

Live on the global edge in 4 seconds. Zero cold starts. V8 sandbox per request. Instant rollback.

$vurb deploy

Building TypeScript bundle…

Sealing V8 Isolate Sandbox…

Deploying to Global Edge…

5. DEBUG IN PRODUCTION

Debug in production

The Inspector shows every tool call, every token consumed, every byte transferred — in real-time. Not in logs. Not in your head.

Inspector — Live Feed

14:32:01→get-user1,240 tok

14:32:02←200 OK342 B

14:32:03→list-orders890 tok

6. CONNECT YOUR AI

Connect your AI client

Copy the MCP endpoint. Paste it into Claude, Cursor, ChatGPT, or VS Code. Your server is live — ask a question and watch the tools fire.

my-server

https://cloud.vinkius.com/mcp/my-server

Claude Desktopready

Cursor IDEready

ChatGPTready

VS Codeready

PATH_03:SKILLS

You have expertise to share. Here's how your knowledge becomes an AI capability.

THE KNOWLEDGE PATH

Write. Publish.
Share.

STEP 01

Open the editor

STEP 02

Write what you know

Give it a name, a description, and write your knowledge in Markdown — the same f…

STEP 03

Publish instantly

Click 'Save'. Your knowledge is immediately available to AI assistants. There's …

STEP 04

Share or sell

Share the link privately with your team or clients. Or list it on the Marketplac…

"You wrote it. AI uses it. That's the whole workflow."

Write → Publish → Share.

1. OPEN THE EDITOR

Open the editor

Immersive Editor

Metadata

Content

2. WRITE WHAT YOU

Write what you know

Give it a name, a description, and write your knowledge in Markdown — the same format you'd use in a document or a note.

.mdSKILL.md

---name: Flat Roof Waterproofingdescription: Expert guidance…---## When to use liquid membraneLiquid membrane is ideal for…

3. PUBLISH INSTANTLY

Publish instantly

Click 'Save'. Your knowledge is immediately available to AI assistants. There's no waiting period, no review queue, no approval process.

Write✓

→

Save✓

→

Published✓

4. SHARE OR SELL

Share or sell

Share the link privately with your team or clients. Or list it on the Marketplace — set a price, earn on every subscription.

MarketplaceListed

Flat Roof WaterproofingExpert guidance on techniques…

$4.99/mo•12 subscribers

SELLER ACCOUNT

Track your earnings

Your Seller Account shows listings, subscribers, MRR and ratings — connected to Stripe for payouts.

Listings

Subscribers

MRR

$139.72

Avg Rating

4.8 ★

Stripe ✓ · Payouts enabled

YOUR LISTINGS

Manage your listings

Each listing has its own cover, price, subscriber count and status. Set any listing to free or paid.

Flat Roof Waterproofing12 subs · $4.99/mo

Active

Solar Panel Maintenance8 subs · $9.99/mo

Active

HVAC Diagnostics8 subs · FREE

Active

CONVERGENCE

Three starting points.
One destination.

API Path

Code Path

Knowledge Path

RUNTIME

Vinkius Cloud

SECURITY_STACK:ACTIVE

V8 Isolate Sandbox✓

Each server runs in an isolated V8 context

Zero-Trust DLP✓

Data redaction before it reaches the model

HMAC Lockfile✓

Tamper-proof integrity for every deployment

Stateful Sessions✓

Persistent context across requests

→

OUTPUT

Any AI Client

CLIENTS:CONNECTED

Claude Desktop & Code

Cursor IDE

ChatGPT

VS Code + Copilot

TRUST_LAYER

"Three starting points. One governance layer. Every AI client."

YOUR TURN

Now you know how. Start building.

Connect an API, write code, or share your expertise — production-ready in 30 seconds.

Start with Google Start with GitHub

Try for Free·No credit card

No credit card required · Free tier available

From where you areto where AI meets you.

Any system. Any expertise. One click.

Paste. Configure.Done.

Find your OpenAPI spec

Paste the URL

Configure security

Deploy

Find your OpenAPI spec

Paste the URL

Configure security

Deploy

Connect your AI assistant

Monitor and govern

Create. Code.Deploy.

Scaffold

Write your tools

Add middleware

Deploy

Scaffold

Write your tools

Add middleware

Deploy

Debug in production

Connect your AI client

Write. Publish.Share.

Open the editor

Write what you know

Publish instantly

Share or sell

Open the editor

Write what you know

Publish instantly

Share or sell

Track your earnings

Manage your listings

Three starting points.One destination.

Vinkius Cloud

Any AI Client

Now you know how. Start building.

Paste. Configure.
Done.

Create. Code.
Deploy.

Write. Publish.
Share.

Three starting points.
One destination.