VSL is available on mobile or desktop

This larger tablet screen size is not currently supported. Please use VSL on mobile, desktop, or iPad mini.

Notification

VSL AI Connections

Your business,
in your AI assistant.

Connect VSL to a compatible AI assistant to read customer enquiries, explore products and prepare your next commerce action.
Choose a plugin ZIP or connect directly with the MCP URL.

Show me how Try a prompt
01 · Choose your AI

Choose your AI. Then choose how to connect.

Select your assistant, then open the ZIP or MCP URL instructions. You only need one connection method.

Connect with ChatGPT

Let’s connect your business.

Connect your VSL business to ChatGPT. Start with your VSL App, or choose a direct MCP connection for your setup.

Option 1 · Download a ZIPLink your VSL App, or connect ChatGPT on desktop.

Bring your business into your chat. Choose the VSL App package for supported web and desktop chats, or the direct MCP package for a desktop assistant.

DownloadsChoose one ZIP, then follow the steps below.

VSL App for ChatGPT

Web + desktop

Products, customers and orders — in your chat.

Already have VSL App access? Choose this ZIP.
Download ZIP

ChatGPT on desktop

Direct MCP

Connect directly with VSL’s MCP tools.

Download ZIP
  1. Choose your connection

    Choose the VSL App ZIP if your account has access to the VSL App. For a direct desktop MCP connection, choose the desktop ZIP.

  2. Upload the selected plugin

    Open Plugins → Add → Upload plugin where available. Select the plugin ZIP directly, review it and finish. A website update ZIP cannot be installed here.

  3. Open your web or desktop connection

    For the App-linked package, select the connected VSL App in a supported chat or use Try in chat when shown. For the desktop package, click Open in desktop app when shown and enable VSL there.

  4. Sign in, then try a prompt

    Complete any VSL connection or OAuth prompt and approve access on VSL. Start a chat or task with VSL selected and try the read-only prompt below. Uploading the ZIP alone does not connect your account.

Option 2 · Use the MCP URLFor assistants with a remote MCP or custom connector setting.

Copy the address from the connection card into your assistant’s server URL field. Choose OAuth to connect your own VSL account.

  1. Add a custom MCP connection

    In ChatGPT Plugins, open Add → Add custom MCP server when available. Name the connection VSL Market. If your VSL App already works, keep it and use Try in chat.

  2. Paste your connection address

    Use the address shown here as the server URL. Choose OAuth so you can sign in.

  3. Sign in and choose access

    When the VSL sign-in page opens, sign in and approve the access you want to give.

  4. Use VSL in your chat

    Open the connected VSL App and select Try in chat where available, or select VSL from the chat’s app controls. Try the read-only prompt below.

What’s inside the VSL package?

The package contains connection configuration and guidance for VSL messages and commerce. It helps your assistant identify the right seller or customer, use current product facts, clarify local wording and follow VSL’s existing confirmation requirements.

The MCP URL connects directly to VSL’s tools. The ZIP also includes workflow guidance. Both use the same VSL server and approved account access. Installation does not retrain your AI or confirm that your account is connected.

More setup help for ChatGPT
Connect with Claude

Let’s connect your business.

Bring your business into a Claude conversation.

Option 1 · Download a ZIPUpload in Claude’s plugin controls, then authorize VSL.

Upload this ZIP through Claude’s custom plugin controls where available. Then complete the VSL connector sign-in. A skill without connected VSL tools cannot access your live account.

DownloadsChoose one ZIP, then follow the steps below.

VSL in Claude

Plugin

Connect VSL, then start a conversation.

Download ZIP
  1. Download the Claude plugin

    Save vsl-claude-plugin.zip. Use this file in the plugin uploader, not as an ordinary chat attachment.

  2. Add it in Customize → Plugins

    Open Claude’s Customize menu and Plugins section. Use the custom upload option if available, review the ZIP and add VSL Market.

  3. Authorize the VSL connection

    Follow any connector sign-in prompt. If VSL tools are unavailable, use the MCP URL instructions to add a custom connector and authorize it.

  4. Select VSL and test

    Start a conversation with the VSL skill selected and try the read-only prompt below. For local Claude Code, extract into a folder named vsl-market and start claude --plugin-dir ./vsl-market from its parent folder, then inspect /mcp.

Option 2 · Use the MCP URLFor assistants with a remote MCP or custom connector setting.

Copy the address from the connection card into your assistant’s server URL field. Choose OAuth to connect your own VSL account.

  1. Add a custom connector

    In a Claude version that supports remote MCP connectors, add one named VSL Market.

  2. Paste your connection address

    Use the address shown here as the remote MCP URL. Let Claude find the sign-in settings.

  3. Sign in and choose access

    Follow the VSL sign-in page and approve the access you want to give.

  4. Turn it on in your chat

    Enable the connector for your chat if asked. Then try a prompt below.

What’s inside the VSL package?

The package contains connection configuration and guidance for VSL messages and commerce. It helps your assistant identify the right seller or customer, use current product facts, clarify local wording and follow VSL’s existing confirmation requirements.

The MCP URL connects directly to VSL’s tools. The ZIP also includes workflow guidance. Both use the same VSL server and approved account access. Installation does not retrain your AI or confirm that your account is connected.

More setup help for Claude
Connect with Gemini

Let’s connect your business.

Use a Gemini app or agent that supports remote MCP tools.

Option 1 · Download a ZIPGemini CLI only: extract the ZIP and install the extension.

For Gemini CLI on your own computer. Extract this ZIP and install the extension from your terminal. The regular Gemini chat app does not install this package.

DownloadsChoose one ZIP, then follow the steps below.

Gemini CLI

Extension

Extract and install with Gemini CLI.

Download ZIP
  1. Download and extract

    Save vsl-gemini-cli-extension.zip and extract the vsl-market folder. Gemini CLI must already be installed.

  2. Install on your own computer

    From the folder containing vsl-market, run: gemini extensions install ./vsl-market. Review and complete installation.

  3. Restart and authorize

    Restart Gemini CLI and inspect /mcp. If authentication is needed, run /mcp auth vsl-market and finish sign-in on VSL.

  4. Check VSL is available

    Use /extensions list and /mcp, then try the read-only prompt below. These commands belong in your Gemini CLI setup, not in VSL’s hosting terminal.

Option 2 · Use the MCP URLFor assistants with a remote MCP or custom connector setting.

Copy the address from the connection card into your assistant’s server URL field. Choose OAuth to connect your own VSL account.

  1. Check for remote tools

    Your Gemini app or agent must support remote MCP tools and OAuth sign-in. The options vary by app.

  2. Add your connection address

    Add a remote tool named VSL Market. Paste the address shown here as the MCP URL.

  3. Sign in and choose access

    Choose OAuth if supported. Follow the VSL sign-in page and approve access.

  4. Ask a simple question

    Try a read-only prompt below. If your app has no connection option, use the full setup guide for your developer.

What’s inside the VSL package?

The package contains connection configuration and guidance for VSL messages and commerce. It helps your assistant identify the right seller or customer, use current product facts, clarify local wording and follow VSL’s existing confirmation requirements.

The MCP URL connects directly to VSL’s tools. The ZIP also includes workflow guidance. Both use the same VSL server and approved account access. Installation does not retrain your AI or confirm that your account is connected.

More setup help for Gemini
Connect with Other AI tools

Let’s connect your business.

For assistants with a remote MCP connection option.

Option 1 · Download a ZIPExtract this reference kit and follow your client’s MCP setup.

This ZIP is a manual setup reference, not an automatic installer. Use your assistant’s own MCP and instructions controls. It includes VSL guidance, reference configuration and the VSL icon.

DownloadsChoose one ZIP, then follow the steps below.

Your AI assistant

Setup kit

Your connection address and commerce guide.

Download ZIP
  1. Download and extract the kit

    Save vsl-mcp-setup-kit.zip. Read README.md and VSL-COMMERCE-GUIDANCE.md.

  2. Add VSL in your client

    Use the client’s remote MCP setting and the connection address here. Treat mcp.json as a reference; preserve your other connections.

  3. Sign in and approve access

    Choose OAuth where supported and follow VSL’s hosted sign-in and access approval.

  4. Add guidance and test

    Load the guidance through your client’s instructions or skill feature. Confirm VSL tools are available, then try the read-only prompt below.

Option 2 · Use the MCP URLFor assistants with a remote MCP or custom connector setting.

Copy the address from the connection card into your assistant’s server URL field. Choose OAuth to connect your own VSL account.

  1. Find the connection option

    Look for a remote MCP server or external tool setting in your assistant.

  2. Paste your connection address

    Use the address shown here as the server URL. Choose OAuth, not API key or password.

  3. Sign in and choose access

    Sign in on the VSL page that opens. Approve only the access you want to give.

  4. Check that it is connected

    Return to your assistant and try a read-only prompt below.

What’s inside the VSL package?

The package contains connection configuration and guidance for VSL messages and commerce. It helps your assistant identify the right seller or customer, use current product facts, clarify local wording and follow VSL’s existing confirmation requirements.

The MCP URL connects directly to VSL’s tools. The ZIP also includes workflow guidance. Both use the same VSL server and approved account access. Installation does not retrain your AI or confirm that your account is connected.

More setup help for Other AI tools
02 · Give it a try

One little prompt. A useful next step.

Connect first. Copy a prompt, then paste it into your AI chat.

Read only

Catch up in one question

Use VSL to show my unread customer enquiries. Do not send anything.

Read only

See what you’re selling

Use VSL to show my current catalog and public Commerce Passport.

Read only

Know who needs you

Use VSL to show customers waiting for a reply.

Draft only

Get a reply ready

Prepare a reply to my latest customer saying the item is available. Do not send it.

Read only

Check unpaid requests

Find accepted requests that remain unpaid.

Read only

Get the short version

Use VSL to summarize my unread customer enquiries. Do not send anything.

A little help, when you need it.Setup answers, everyday examples and developer details. Open just the topic you need.
Connection basicsThe address, first steps and setup checklist.
AI connection guide

Start with the essentials

Choose the ZIP or MCP URL instructions above. With either option, authorize your account on VSL. The sections below explain what to check next.

https://vsl.market/dev/messages/ai-connect/mcp
Start here

From connection to your first VSL question

Most users only need the MCP URL and their normal VSL sign-in. Follow these stages in order.

ConnectAdd https://vsl.market/dev/messages/ai-connect/mcp as the MCP server or external-tool URL and choose OAuth.
Authorize on VSLLet your AI open the VSL sign-in page. Enter your normal VSL email/username and password there only.
Verify the connectionReturn to your AI client and confirm VSL is connected before requesting any actions.
Start with a readAsk: “Use VSL to show my unread customer enquiries. Do not send anything.”
What happens during authorization?PKCE, scopes and tokens

The client discovers VSL OAuth metadata, registers a public client when required, opens VSL's authorization page, uses PKCE S256, and exchanges the approved authorization code for a resource-bound bearer token.

Supported clients manage the access and refresh tokens themselves. You should not manually copy tokens into ordinary AI settings.

Keep VSL credentials on VSL. Never type your VSL password, access token or refresh token into an AI prompt or normal client settings.

Any supported MCP client

Use this route when your assistant supports a remote MCP server. Everything needed for the connection is shown here.

Serverhttps://vsl.market/dev/messages/ai-connect/mcp AuthenticationOAuth Sign-inOn VSL
Connect a generic MCP clientUse these values in the client's connection form
Add the serverPaste the VSL MCP URL into the remote MCP/server field.
Choose OAuthDo not select API key, Basic Auth or password authentication.
Sign in on VSLApprove access on the VSL-hosted authorization page.
Test the connectionAsk a read-only VSL question first.
OAuth configuration for this clientOpen only if the client asks for manual OAuth fields
Server / MCP URLhttps://vsl.market/dev/messages/ai-connect/mcp
AuthenticationOAuth 2.1 authorization code + PKCE S256
Authorization server / issuerhttps://vsl.market/dev/messages/ai-connect
Authorization URLhttps://vsl.market/dev/messages/ai-connect/oauth/authorize
Token URLhttps://vsl.market/dev/messages/ai-connect/oauth/token
Registration URLhttps://vsl.market/dev/messages/ai-connect/oauth/register
Resourcehttps://vsl.market/dev/messages/ai-connect/mcp
Client secretLeave blank for a public PKCE client
OIDCDisabled
Help for your assistantExtra setup details for ChatGPT, Claude and Gemini.

ChatGPT

Create the VSL connection in ChatGPT and use the values below if Advanced OAuth settings do not populate automatically.

NameVSL Market Server URLhttps://vsl.market/dev/messages/ai-connect/mcp AuthenticationOAuth
ChatGPT setupWhat to enter and what happens next
Create a new app/connectionName it VSL Market and paste the VSL MCP server URL.
Select OAuthLet ChatGPT discover VSL OAuth automatically first.
Create and authorizeWhen VSL opens, sign in with your normal VSL account and approve access.
Return to ChatGPTConfirm the VSL app is connected, then enable it for the chat if required.
ChatGPT Advanced OAuth settingsExact values for the fields in ChatGPT
Server URLhttps://vsl.market/dev/messages/ai-connect/mcp
AuthenticationOAuth
Base scopesLeave blank unless your workspace explicitly requires a value
Auth URLhttps://vsl.market/dev/messages/ai-connect/oauth/authorize
Token URLhttps://vsl.market/dev/messages/ai-connect/oauth/token
Registration URLhttps://vsl.market/dev/messages/ai-connect/oauth/register
Authorization server basehttps://vsl.market/dev/messages/ai-connect
Resourcehttps://vsl.market/dev/messages/ai-connect/mcp
OIDC enabledNo — leave disabled
Client ID / secretLet VSL dynamic registration create the client ID; do not enter a client secret for the public PKCE flow

If ChatGPT has already discovered these values, do not replace them manually.

Start using VSL in ChatGPTRecommended first requests

“Use VSL to show my unread customer enquiries. Do not send anything.”

“Use VSL to show my current catalog and public Commerce Passport.”

“Prepare a reply to my latest customer saying the item is available. Do not send it.”

Claude

Use the VSL MCP URL in a Claude surface that supports remote MCP or custom connectors. The complete OAuth fallback is included here.

NameVSL Market Remote MCP URLhttps://vsl.market/dev/messages/ai-connect/mcp AuthenticationOAuth
Claude setupRemote MCP / custom connector
Add VSL MarketCreate a remote MCP/custom connector and paste the VSL MCP URL.
Use OAuthAllow Claude to discover VSL OAuth where supported.
Authorize on VSLSign in and approve the permissions on VSL.
Enable VSL for the chatIf Claude uses per-chat tools, enable the connector before asking a VSL question.
Claude OAuth configurationUse only when Claude exposes manual OAuth fields
MCP serverhttps://vsl.market/dev/messages/ai-connect/mcp
Authorization server / issuerhttps://vsl.market/dev/messages/ai-connect
Authorization URLhttps://vsl.market/dev/messages/ai-connect/oauth/authorize
Token URLhttps://vsl.market/dev/messages/ai-connect/oauth/token
Registration URLhttps://vsl.market/dev/messages/ai-connect/oauth/register
Resourcehttps://vsl.market/dev/messages/ai-connect/mcp
PKCES256
Client secretLeave blank for a public PKCE client
Start using VSL in ClaudeRead before write

“Use VSL to summarize my unread customer enquiries. Do not send anything.”

Gemini and Google-based agents

Google products do not all expose the same MCP form. Use the values below in the Gemini surface or agent runtime that supports remote MCP/custom tools.

NameVSL Market MCP URLhttps://vsl.market/dev/messages/ai-connect/mcp AuthenticationOAuth when supported
Gemini remote MCP setupFor a surface that exposes remote MCP/custom tools
Add a remote toolName it VSL Market and paste the MCP URL.
Choose OAuthIf the surface supports OAuth discovery, let it discover VSL first.
Authorize on VSLComplete sign-in and consent only on the VSL-hosted page.
Test VSLStart with a read-only catalog or enquiry request.
Gemini OAuth configurationUse when your Gemini surface exposes manual OAuth fields
MCP URLhttps://vsl.market/dev/messages/ai-connect/mcp
Authorization server / issuerhttps://vsl.market/dev/messages/ai-connect
Authorization URLhttps://vsl.market/dev/messages/ai-connect/oauth/authorize
Token URLhttps://vsl.market/dev/messages/ai-connect/oauth/token
Registration URLhttps://vsl.market/dev/messages/ai-connect/oauth/register
Resourcehttps://vsl.market/dev/messages/ai-connect/mcp
PKCES256
Client secretLeave blank for a public PKCE client
Gemini API / server-side agentsDeveloper runtime

Use the MCP server name and URL over Streamable HTTP. If your runtime does not perform VSL OAuth automatically, complete OAuth securely in your server-side application and attach the bearer token there. Do not expose tokens in browser code, prompts or source control.

{ "type": "mcp_server", "name": "vsl_market", "url": "https://vsl.market/dev/messages/ai-connect/mcp" }
Sign-in & permissionsOAuth discovery and the access you approve.
Reference only

VSL OAuth reference

You normally do not need this section because the relevant values are included inside each AI assistant guide above. This anchor remains available for developers and saved links.

Discovery metadata URLsAdvanced troubleshooting and development
Protected-resource metadatahttps://vsl.market/.well-known/oauth-protected-resource/dev/messages/ai-connect/mcp
Authorization-server metadatahttps://vsl.market/.well-known/oauth-authorization-server/dev/messages/ai-connect
What can I do next?Prompts, product discovery and incoming AI requests.
After configuration

How to start using VSL from your assistant

You can ask in normal language. You do not need to type tool names, endpoint paths or OAuth details after the connection is working.

Ask normally“Use VSL to show customers waiting for a reply.”
VSL checks accessThe token, resource, scope and account ownership are checked before data is returned.
Reads can return directlyAuthorized catalog, conversations, requests and other permitted VSL context can be returned.
Writes remain controlledReplies, order changes and payment-request actions can require preparation and confirmation.

Help AI find your store

Your public business and product information can be discoverable independently of connecting your own assistant.

Commerce PassportPublic business identity and capabilities.
VSL CatalogActive products, prices and descriptions.
Public discovery endpointStructured product discovery
https://vsl.market/dev/messages/ai-connect/commerce/v1/discovery

How AI orders work

A customer's AI can send a structured request, but you remain in control and payment truth remains inside VSL.

Customer searchesAI sends requestYou decideCustomer pays on VSL
What you can do with an incoming AI requestAccept, counter or ask for information
  • Accept: create an official VSL order draft.
  • Counteroffer: propose a revised price, quantity or delivery fee.
  • Ask a question: clarify size, color or delivery requirements.
  • Request payment: generate an official VSL checkout link after the commercial details are agreed.

Requests appear in your Agentic Commerce Inbox.

For developersDiscovery filters, API examples and test prompts.

Agentic commerce discovery parameters

Production discovery is narrow and deterministic. Filters are strict when supplied.

View discovery parametersquery, country, currency, price range, result type, limit and cursor
ParameterProduction rule
queryRequired plain-language search text. Maximum 160 characters.
countryOptional two-letter ISO 3166-1 alpha-2 code. Strict merchant-country filter.
currencyOptional three-letter ISO 4217 code. Strict product-currency filter.
min_priceOptional non-negative minimum listed product price.
max_priceOptional non-negative maximum listed product price.
result_typeOptional all, product, or merchant filter. Defaults to all.
limitOptional page size. Default 10; minimum 1; maximum 25.
cursorOpaque next_cursor; reuse with the same query and all supplied filters.

Endpoint: https://vsl.market/dev/messages/ai-connect/commerce/v1/discovery.

Developers and custom agents

Use the Business API when you are building your own server-side client or automation instead of a normal MCP-capable assistant.

POST https://vsl.market/dev/messages/ai-connect/api
Example API requestBearer token + idempotency key
POST https://vsl.market/dev/messages/ai-connect/api HTTP/1.1 Host: vsl.marketAuthorization: Bearer YOUR_ACCESS_TOKEN Content-Type: application/json Idempotency-Key: 550e8400-e29b-41d4-a716-446655440000 { "tool": "get_unread_conversations", "arguments": { "limit": 20 } }
Developer rulesTransport, permissions and prepared actions
  • Transport: Streamable HTTP over public HTTPS.
  • Authorization: OAuth 2.1 authorization code + PKCE S256 with resource-bound bearer tokens.
  • Permissions: scopes fail closed and remain tied to the approved connection.
  • References: VSL references are opaque; internal database IDs are not part of the public contract.
  • Prepared actions: sensitive writes can require preview and seller confirmation before execution.

Test the connection

Start with reads, then move to prepared actions once you know the connection and permissions are correct.

Read-only

“Show my unread sales messages on VSL.”

“Which customer conversations need my attention today?”

Prepared actions

“Prepare a reply to my latest customer saying the item is available. Do not send it.”

“Find accepted requests that remain unpaid.”

VSL remains your sales system of record.No payouts, withdrawals or refunds are exposed through this connector.