# Connect border.bot to your AI tools

> Connect border.bot to Claude, ChatGPT, Codex, Cursor, Visual Studio Code, Gemini and other MCP clients. Classify products and calculate landed cost from chat.

Source: https://border.bot/mcp
Last updated: 2026-10-09

Our remote MCP server lets AI assistants and coding agents classify products and calculate landed cost for you. Add one URL, sign in with your border.bot account, and choose which workspace pays.

## Server URL

```text
https://api.border.bot/mcp
```

Streamable HTTP transport with OAuth 2.1. Your client finds the sign-in endpoints itself, so there is no API key to paste.

## Set up your client

Steps for 34 clients, checked against each vendor’s own documentation on 7 October 2026. Menu names change between versions and plans; if one has moved, look for “connectors”, “integrations” or “MCP” in your client’s settings.

- Assistants: [Claude](https://border.bot/mcp#claude), [ChatGPT](https://border.bot/mcp#chatgpt), [Gemini](https://border.bot/mcp#gemini), [Perplexity](https://border.bot/mcp#perplexity), [Vibe Work (formerly Le Chat)](https://border.bot/mcp#mistral-vibe), [Raycast](https://border.bot/mcp#raycast), [LM Studio](https://border.bot/mcp#lm-studio)
- Editors and IDEs: [Cursor](https://border.bot/mcp#cursor), [Visual Studio Code](https://border.bot/mcp#vscode), [Devin Desktop (formerly Windsurf)](https://border.bot/mcp#devin-desktop), [Visual Studio](https://border.bot/mcp#visual-studio), [Cline](https://border.bot/mcp#cline), [Zed](https://border.bot/mcp#zed), [Kiro](https://border.bot/mcp#kiro), [Replit](https://border.bot/mcp#replit), [v0](https://border.bot/mcp#v0), [JetBrains Air](https://border.bot/mcp#jetbrains-air), [TraeCode](https://border.bot/mcp#trae)
- Command line: [Claude Code](https://border.bot/mcp#claude-code), [Codex](https://border.bot/mcp#codex), [Gemini CLI](https://border.bot/mcp#gemini-cli), [GitHub Copilot CLI](https://border.bot/mcp#github-copilot-cli), [OpenCode](https://border.bot/mcp#opencode), [Warp](https://border.bot/mcp#warp), [Junie CLI](https://border.bot/mcp#junie), [Amp](https://border.bot/mcp#amp), [goose](https://border.bot/mcp#goose), [Qwen Code](https://border.bot/mcp#qwen-code), [Kimi Code CLI](https://border.bot/mcp#kimi-cli)
- Automation: [Microsoft Copilot Studio](https://border.bot/mcp#copilot-studio), [n8n](https://border.bot/mcp#n8n), [Zapier](https://border.bot/mcp#zapier), [Make](https://border.bot/mcp#make), [Postman](https://border.bot/mcp#postman)

Clients that aren’t listed can connect too, as long as they support remote MCP servers with OAuth. Point yours at the server URL; it finds the authorization server and opens the border.bot sign-in page, where you approve the connection and choose the workspace it may bill.

## Assistants

Chat apps where you add border.bot once and then ask for a code or a landed cost in any conversation.

### Claude

By Anthropic. Free, Pro, Max, Team and Enterprise (Free is limited to one custom connector). On Team and Enterprise an Owner adds it for the organization.

1. Open Customize › Connectors (claude.ai/customize/connectors), click “+ Add”, then “Add custom connector”.
2. Enter border.bot as the name and `https://api.border.bot/mcp` as the remote MCP server URL, then click “Continue”.
3. Review the authentication settings Claude detected and click “Continue”; under Authentication choose “Sign in now” and under OAuth client “Use Claude’s published identity (Recommended)”, then click “Add”.
4. Sign in to border.bot and approve access in the window that opens.
5. In a chat, open “+” › “Connectors” and switch border.bot on.

Official setup guide: https://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp

Notes: One connector covers claude.ai, Claude Desktop, Cowork and the mobile apps. Claude connects from Anthropic’s cloud, so the URL must be publicly reachable. Team/Enterprise: an Owner adds it under Organization settings › Connectors (“Add” › “Custom” › “Web”), then each member clicks “Connect” in Customize › Connectors.

### ChatGPT

By OpenAI. ChatGPT on the web. OpenAI’s help center lists full MCP (read and write tools) for Business, Enterprise and Edu (beta) and read/fetch-only custom MCP for Pro in developer mode; workspace admins control who may create plugins with MCP servers.

1. Go to ChatGPT Plugins (chatgpt.com/plugins), select the plus button, then “Add custom MCP server”.
2. Enter border.bot as the name and, under “Connection”, `https://api.border.bot/mcp` as the “Server URL”.
3. Choose “OAuth” for authentication, review the risk warning, select “I understand and want to continue”, then “Create as a plugin”.
4. Open the new plugin in your personal plugins, install it with the plus button and sign in to border.bot when asked.
5. In a new chat, type @ and select border.bot.

Official setup guide: https://developers.openai.com/api/docs/guides/custom-mcp-server

Notes: Not available in the mobile apps. ChatGPT supports Client ID Metadata Documents and Dynamic Client Registration. Write tools ask for confirmation unless marked read-only. Workspaces still on the older flow create it under Settings › Apps › Create with developer mode on.

### Gemini

By Google. Personal Google Accounts only (not work or school), 18+ in the US, with Keep Activity on; English only. Added on the web, then usable in the web and mobile apps.

1. On a computer, go to gemini.google.com and click Settings › “Connected Apps” at the bottom.
2. Under “Custom apps”, click “Add a custom app” and enter `https://api.border.bot/mcp` as the MCP server URL.
3. Click “Next” and follow the on-screen instructions to sign in to border.bot.
4. In a chat, enter @ and select the app to make sure Gemini uses it.

Official setup guide: https://support.google.com/gemini/answer/17209137?hl=en

Notes: Gemini registers itself with Dynamic Client Registration, so no client credentials are needed. Write actions always need manual confirmation.

### Perplexity

Pro, Max and Enterprise. Enterprise members can add their own only when an admin allows custom connectors.

1. Go to Account settings › Connectors, click “+ Custom connector” in the top-right corner and select “Remote”.
2. Name: border.bot; MCP Server URL: `https://api.border.bot/mcp`; Authentication: “OAuth”; Transport: “Streamable HTTP”.
3. Check the acknowledgement box and click “Add”.
4. Click the border.bot connector card to start the sign-in and enable it.

Official setup guide: https://www.perplexity.ai/help-center/en/articles/13915507-adding-custom-remote-connectors

Notes: Perplexity discovers border.bot’s OAuth endpoints and registers itself dynamically, so no client ID or secret is needed. Enterprise admins can add it for everyone under Enterprise settings › Permissions › Connectors permissions.

### Vibe Work (formerly Le Chat)

By Mistral AI. Administrators only: on Free, Pro and Student plans the account owner is the administrator.

1. In Vibe Work (chat.mistral.ai), open the “Connectors” page from the sidebar.
2. Click “+ Add Connector” and switch to the “Custom MCP Connector” tab.
3. Connector name: `borderbot`; Server URL: `https://api.border.bot/mcp`; then click “Connect”.
4. Complete the OAuth consent flow when prompted.

Official setup guide: https://docs.mistral.ai/vibe/work/connectors/mcp-connectors

Notes: Formerly Le Chat. The authentication method (OAuth 2.1 with dynamic client registration) is detected automatically. Under “Connectors” › “My Connectors” › “Functions” you can mark read-only tools “Always allow”.

### Raycast

Raycast Pro.

1. Open the “Install MCP Server” command in Raycast.
2. Name: border.bot; Transport: “HTTP”; URL: `https://api.border.bot/mcp`; OAuth Type: “Dynamic”.
3. Press “Install MCP Server” (⌘↵), then select “Sign In” to authorize border.bot.
4. In AI Chat or Quick AI, type @ and the server’s name to use it.

Official setup guide: https://manual.raycast.com/ai/model-context-protocol

Notes: Works on desktop and iOS (HTTP servers only on iOS). The “Logout Server” action in “Manage MCP Servers” clears saved tokens.

### LM Studio

By Element Labs.

1. Click “Add to LM Studio”, or open the “Program” tab in the right-hand sidebar, click “Install” › “Edit mcp.json” and add the config snippet.
2. LM Studio opens border.bot’s authorization page in your browser; approve access.
3. border.bot’s tools are then available to your models in chat.

Config file mcp.json (Program › Install › Edit mcp.json):

```json
{
  "mcpServers": {
    "borderbot": {
      "url": "https://api.border.bot/mcp"
    }
  }
}
```

One-click install: [Add to LM Studio](lmstudio://add_mcp?name=borderbot&config=eyJ1cmwiOiJodHRwczovL2FwaS5ib3JkZXIuYm90L21jcCJ9)

Official setup guide: https://lmstudio.ai/docs/integrations/mcp-remote

Notes: OAuth sign-in for MCP servers needs LM Studio 0.4.10 or later. Install-link format: lmstudio.ai/docs/app/mcp/deeplink.

## Editors and IDEs

Code editors and app builders whose agents can call border.bot while they work on your catalogue or checkout code.

### Cursor

By Anysphere.

1. Click “Add to Cursor” and confirm when Cursor prompts to install the server, or add the config snippet to `~/.cursor/mcp.json`.
2. Open “Customize” in the sidebar, find `borderbot` and complete the OAuth sign-in it asks for (Cursor CLI: `agent mcp login borderbot`).
3. Ask Agent to use border.bot; Cursor asks for approval before running MCP tools.

Config file `~/.cursor/mcp.json`:

```json
{
  "mcpServers": {
    "borderbot": {
      "url": "https://api.border.bot/mcp"
    }
  }
}
```

One-click install: [Add to Cursor](cursor://anysphere.cursor-deeplink/mcp/install?name=borderbot&config=eyJ1cmwiOiJodHRwczovL2FwaS5ib3JkZXIuYm90L21jcCJ9)

Official setup guide: https://cursor.com/docs/mcp

Notes: The same `mcp.json` configures the Cursor CLI (`agent`). Install-link format: cursor.com/docs/mcp/install-links.

### Visual Studio Code

By Microsoft. Uses GitHub Copilot. Copilot Business and Enterprise members need the “MCP servers in Copilot” policy enabled by their organization.

1. Click “Install in VS Code” and confirm in VS Code, or run “MCP: Add Server” from the Command Palette, choose the HTTP type, enter `https://api.border.bot/mcp` and the name `borderbot`, and pick “Copilot Global” to use it in every workspace.
2. Trust the server when prompted; on first connection VS Code opens a browser window to sign in to border.bot.
3. Open the Chat view and ask for border.bot; select “Configure Tools” in the chat input to see its tools.

Command:

```bash
code --add-mcp '{"name":"borderbot","type":"http","url":"https://api.border.bot/mcp"}'
```

Config file `~/.copilot/mcp-config.json`:

```json
{
  "mcpServers": {
    "borderbot": {
      "type": "http",
      "url": "https://api.border.bot/mcp",
      "tools": [
        "*"
      ]
    }
  }
}
```

One-click install: [Install in VS Code](vscode:mcp/install?%7B%22name%22%3A%22borderbot%22%2C%22type%22%3A%22http%22%2C%22url%22%3A%22https%3A%2F%2Fapi.border.bot%2Fmcp%22%7D)

Official setup guide: https://code.visualstudio.com/docs/agent-customization/mcp-servers

Notes: `~/.copilot/mcp-config.json` is the portable user format VS Code now prefers (shared with GitHub Copilot CLI). The older VS Code format (`.vscode/mcp.json`, or the user-profile file from “MCP: Open User Configuration”) uses a top-level "servers" object: `{"servers": {"borderbot": {"type": "http", "url": "https://api.border.bot/mcp"}}}`. For VS Code Insiders use the `vscode-insiders:` scheme.

### Devin Desktop (formerly Windsurf)

By Cognition. Enterprise users must turn MCP on in settings.

1. Open the MCP config: in the Cascade panel click “..” (Actions) › “Open MCP config file” (`~/.config/devin/mcp_config.json`), or run `devin mcp add -s user borderbot https://api.border.bot/mcp` with Devin CLI.
2. Add border.bot under "mcpServers" using the config snippet and save.
3. Sign in to border.bot when prompted on first use; if the server shows “Needs auth”, click “Authenticate”.

Command:

```bash
devin mcp add -s user borderbot https://api.border.bot/mcp
```

Config file `~/.config/devin/mcp_config.json`:

```json
{
  "mcpServers": {
    "borderbot": {
      "url": "https://api.border.bot/mcp"
    }
  }
}
```

Official setup guide: https://docs.devin.ai/desktop/cascade/mcp

Notes: Formerly Windsurf (windsurf.com now redirects to Devin Desktop). The file is shared by the legacy Cascade agent, the default Devin Local agent and Devin CLI (Windows: `%APPDATA%\devin\mcp_config.json`). Devin Local asks before each MCP tool call; `devin mcp login borderbot` re-runs the sign-in.

### Visual Studio

By Microsoft. Visual Studio 2026, or Visual Studio 2022 17.14 and later, with GitHub Copilot. Organization MCP allow-list policies apply.

1. Click “Install in Visual Studio”, or in the Copilot Chat pane switch to Agent mode, select “Tools”, then the plus (+) button › “Add custom MCP server”, and enter `borderbot` and `https://api.border.bot/mcp`.
2. In `.mcp.json`, select “Authentication Required” (or “Manage Authentication”) from the CodeLens above `borderbot` and sign in to border.bot in the browser pop-up.
3. Enable border.bot’s tools in the tool picker and choose “Allow” when Copilot asks to run one.

Config file `%USERPROFILE%\.mcp.json`:

```json
{
  "servers": {
    "borderbot": {
      "url": "https://api.border.bot/mcp"
    }
  }
}
```

One-click install: [Install in Visual Studio](https://vs-open.link/mcp-install?%7B%22name%22%3A%22borderbot%22%2C%22url%22%3A%22https%3A%2F%2Fapi.border.bot%2Fmcp%22%7D)

Official setup guide: https://learn.microsoft.com/en-us/visualstudio/ide/mcp-servers

Notes: Visual Studio supports any MCP-spec OAuth provider. The install-link format is the one Microsoft’s own “Install” buttons on that page use. A solution-level `.mcp.json` works too.

### Cline

1. In the Cline panel, click the “MCP Servers” icon and open the “Remote Servers” tab.
2. Server Name: `borderbot`; Server URL: `https://api.border.bot/mcp`; Transport Type: “Streamable HTTP”; then click “Add Server”.
3. Complete the border.bot authorization in your browser (Cline CLI: run `cline mcp` and choose “Authorize OAuth”).

Command:

```bash
cline mcp add borderbot https://api.border.bot/mcp --transport http --yes
```

Config file `~/.cline/data/settings/cline_mcp_settings.json`:

```json
{
  "mcpServers": {
    "borderbot": {
      "type": "streamableHttp",
      "url": "https://api.border.bot/mcp"
    }
  }
}
```

Official setup guide: https://docs.cline.bot/mcp/mcp-overview

Notes: The path above is the Cline CLI’s; in the IDE extension use “MCP Servers” › “Configure” › “Configure MCP Servers” to edit the same JSON. Keep "type": "streamableHttp", without it Cline uses legacy SSE.

### Zed

By Zed Industries.

1. Open Settings › AI › MCP Servers, click “Add Server” and choose “Add Remote Server”.
2. Enter `borderbot` and `https://api.border.bot/mcp` (no Authorization header) and save.
3. Zed prompts you to authenticate with the standard MCP OAuth flow, sign in to border.bot.
4. When the indicator next to `borderbot` is green (“Server is active”), use it from the Agent Panel.

Config file `~/.config/zed/settings.json`:

```json
{
  "context_servers": {
    "borderbot": {
      "url": "https://api.border.bot/mcp"
    }
  }
}
```

Official setup guide: https://zed.dev/docs/ai/mcp

Notes: Zed also forwards configured MCP servers to external agents over ACP.

### Kiro

By Amazon Web Services.

1. Run “Kiro: Open user MCP config (JSON)” from the Command Palette (or select the “Open MCP Config” icon in the Kiro panel).
2. Add border.bot under "mcpServers" using the config snippet and save, servers reconnect automatically.
3. Kiro opens border.bot’s authorization page; sign in and approve.
4. Check the MCP servers tab in the Kiro panel to confirm it is connected.

Config file `~/.kiro/settings/mcp.json`:

```json
{
  "mcpServers": {
    "borderbot": {
      "url": "https://api.border.bot/mcp"
    }
  }
}
```

Official setup guide: https://kiro.dev/docs/mcp/configuration/

Notes: If MCP is off, enable the MCP support setting (Settings, search “MCP”). Kiro CLI reads the same file. Kiro registers via Dynamic Client Registration and requires the issuer in border.bot’s protected-resource metadata to match the authorization server’s `issuer` exactly.

### Replit

1. Click “Add to Replit”, or go to replit.com/integrations, scroll to “MCP Servers for Replit Agent” and click “+ Add MCP server”.
2. Display name: border.bot; MCP Server URL: `https://api.border.bot/mcp`.
3. Click “Test & save” and complete the border.bot OAuth sign-in Replit walks you through.
4. Mention border.bot in an Agent chat to use its tools in any project.

One-click install: [Add to Replit](https://replit.com/integrations?mcp=eyJkaXNwbGF5TmFtZSI6ImJvcmRlci5ib3QiLCJiYXNlVXJsIjoiaHR0cHM6Ly9hcGkuYm9yZGVyLmJvdC9tY3AifQ%3D%3D)

Official setup guide: https://docs.replit.com/build/connect-via-mcp

### v0

By Vercel.

1. In v0, open the + menu in the prompt form and select “MCPs”.
2. Configure a custom MCP server with `https://api.border.bot/mcp` and choose “OAuth” as the authentication.
3. Sign in to border.bot when prompted; v0 then considers its tools in new and existing chats.

Official setup guide: https://v0.app/docs/MCP

Notes: MCP tools are available to v0 while it generates; the apps v0 builds can’t call them. Tool calls follow your global permission mode (Ask, Auto or Full).

### JetBrains Air

By JetBrains.

1. Open Settings › AI › MCP Servers and make sure “Enable MCP support” is on.
2. Click “Add Global MCP Server”, paste the config snippet and save.
3. Click “Connect” next to `borderbot` and approve the connection on border.bot’s authorization page.

Config file `.air/mcp.json`:

```json
{
  "mcpServers": {
    "borderbot": {
      "url": "https://api.border.bot/mcp"
    }
  }
}
```

Official setup guide: https://www.jetbrains.com/help/air/mcp-servers.html

Notes: `.air/mcp.json` is the project-local location (“Add Local MCP Server”); a global server is stored by the app. Air cloud tasks take MCP servers from the repository’s `mcp.json` or a JetBrains Air Teams connector instead.

### TraeCode

By ByteDance.

1. Click “Add to TraeCode” and confirm the configuration with “Confirm”, or open Settings (top-right icon) › “MCP” › “Add” › “Add Manually”.
2. If adding manually, paste the config snippet and click “Confirm”.
3. Sign in to border.bot if TraeCode prompts you to.

Config file TraeCode mcp.json (Settings › MCP › Add › Add Manually):

```json
{
  "mcpServers": {
    "borderbot": {
      "url": "https://api.border.bot/mcp"
    }
  }
}
```

One-click install: [Add to TraeCode](trae://trae.ai-ide/mcp-import?type=http&name=borderbot&config=eyJ1cmwiOiJodHRwczovL2FwaS5ib3JkZXIuYm90L21jcCJ9)

Official setup guide: https://docs.trae.ai/ide/add-mcp-servers

Notes: TraeCode (the TRAE IDE) documents remote HTTP servers with header authentication only; OAuth sign-in isn’t documented, so it may not connect to border.bot’s OAuth-only endpoint. Install-link format: docs.trae.ai/ide/mcp-server-install-links.

## Command line

Coding agents that run in a terminal. Most add the server with one command.

### Claude Code

By Anthropic.

1. Run `claude mcp add --transport http --scope user borderbot https://api.border.bot/mcp`.
2. Start Claude Code, run `/mcp`, select `borderbot` and follow the steps in your browser to log in (or run `claude mcp login borderbot` from your shell).
3. Run `/mcp` again to confirm `borderbot` is connected.

Command:

```bash
claude mcp add --transport http --scope user borderbot https://api.border.bot/mcp
```

Config file `.mcp.json`:

```json
{
  "mcpServers": {
    "borderbot": {
      "type": "http",
      "url": "https://api.border.bot/mcp"
    }
  }
}
```

Official setup guide: https://code.claude.com/docs/en/mcp

Notes: `--scope user` makes border.bot available in every project; the `.mcp.json` snippet is for sharing it with a team via a repository. Claude Code discovers OAuth and border.bot’s Client ID Metadata Document support automatically and refreshes tokens itself. If you added border.bot as a connector on claude.ai and log in to Claude Code with that account, it already appears in `/mcp`.

### Codex

By OpenAI.

1. CLI: run `codex mcp add borderbot --url https://api.border.bot/mcp`, then `codex mcp login borderbot` and sign in to border.bot in your browser.
2. IDE extension: open the gear menu › “MCP servers” › “Add server”, enter `borderbot`, choose “Streamable HTTP”, paste `https://api.border.bot/mcp`, save and select “Restart extension”, then “Authenticate”.
3. ChatGPT desktop app: “Settings” › “MCP servers” › “Add server” with the same values, save and select “Restart”, then “Authenticate”.
4. Type `/mcp` to see the connected servers.

Command:

```bash
codex mcp add borderbot --url https://api.border.bot/mcp
```

Config file `~/.codex/config.toml`:

```toml
[mcp_servers.borderbot]
url = "https://api.border.bot/mcp"
```

Official setup guide: https://learn.chatgpt.com/docs/extend/mcp

Notes: The Codex CLI, IDE extension and ChatGPT desktop app share `~/.codex/config.toml`, so adding border.bot once covers all three. Codex uses a Client ID Metadata Document when the authorization server advertises it and otherwise Dynamic Client Registration, with a loopback callback on 127.0.0.1.

### Gemini CLI

By Google.

1. Run `gemini mcp add --transport http --scope user borderbot https://api.border.bot/mcp`.
2. Start `gemini` and run `/mcp auth borderbot`; sign in to border.bot in the browser window that opens.
3. Run `/mcp list` to check that `borderbot` is connected.

Command:

```bash
gemini mcp add --transport http --scope user borderbot https://api.border.bot/mcp
```

Config file `~/.gemini/settings.json`:

```json
{
  "mcpServers": {
    "borderbot": {
      "httpUrl": "https://api.border.bot/mcp"
    }
  }
}
```

Official setup guide: https://geminicli.com/docs/tools/mcp-server/

Notes: Use `httpUrl` for Streamable HTTP. `url` means SSE in Gemini CLI. OAuth needs a local browser and a localhost callback, so it doesn’t work in headless or SSH sessions without forwarding.

### GitHub Copilot CLI

By GitHub. All Copilot plans. With Copilot from an organization, the Copilot CLI policy must be enabled, and MCP allowlist policies apply.

1. Run `copilot mcp add --transport http borderbot https://api.border.bot/mcp`, or in interactive mode enter `/mcp add`, pick “HTTP or SSE”, paste the URL and press Ctrl+S.
2. If `borderbot` shows `needs-auth`, run `/mcp auth borderbot` and sign in to border.bot in your browser.
3. Run `/mcp show borderbot` to see its status and tools.

Command:

```bash
copilot mcp add --transport http borderbot https://api.border.bot/mcp
```

Config file `~/.copilot/mcp-config.json`:

```json
{
  "mcpServers": {
    "borderbot": {
      "type": "http",
      "url": "https://api.border.bot/mcp",
      "tools": [
        "*"
      ]
    }
  }
}
```

Official setup guide: https://docs.github.com/en/copilot/how-tos/copilot-cli/customize-copilot/add-mcp-servers

Notes: Copilot CLI registers itself with Dynamic Client Registration. `~/.copilot/mcp-config.json` is also read by Visual Studio Code.

### OpenCode

By Anomaly.

1. Add border.bot under "mcp" in `~/.config/opencode/opencode.json` using the config snippet (or run `opencode mcp add` and choose a remote server).
2. Run `opencode mcp auth borderbot` and authorize in your browser. OpenCode also prompts the first time you use it.
3. Run `opencode mcp list` to check its connection and auth status.

Config file `~/.config/opencode/opencode.json`:

```json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "borderbot": {
      "type": "remote",
      "url": "https://api.border.bot/mcp"
    }
  }
}
```

Official setup guide: https://opencode.ai/docs/mcp-servers/

Notes: OpenCode uses Dynamic Client Registration automatically; no client ID is needed.

### Warp

1. Open Settings › Agents › MCP servers (or run “Open MCP Servers” from the Command Palette) and click “+ Add”.
2. Paste the config snippet and save.
3. Start `borderbot`; Warp opens a browser to sign in to border.bot the first time.

Config file `~/.warp/.mcp.json`:

```json
{
  "mcpServers": {
    "borderbot": {
      "url": "https://api.border.bot/mcp"
    }
  }
}
```

Official setup guide: https://docs.warp.dev/agents/capabilities/mcp/

Notes: Servers in `~/.warp/.mcp.json` start automatically and show an authentication modal on first spawn. Credentials stay on the device, so sign in again on a new machine.

### Junie CLI

By JetBrains.

1. Add the config snippet to `~/.junie/mcp/mcp.json`, or run `/mcp`, press Ctrl+A to add a server with the MCP Installation Assistant, and choose “User scope” and “Remote”.
2. border.bot shows the “Authorization required” status: select it, choose “Authorize” and log in on the border.bot page that opens.
3. Check that its status changes to “Active”.

Config file `~/.junie/mcp/mcp.json`:

```json
{
  "mcpServers": {
    "borderbot": {
      "url": "https://api.border.bot/mcp"
    }
  }
}
```

Official setup guide: https://junie.jetbrains.com/docs/junie-cli-mcp-configuration.html

Notes: Junie in JetBrains IDEs reads the same `~/.junie/mcp/mcp.json`, but JetBrains documents OAuth sign-in only for Junie CLI. JetBrains AI Assistant chat can’t sign in to OAuth MCP servers yet.

### Amp

1. Run `amp mcp add borderbot https://api.border.bot/mcp` (saved to `amp.mcpServers` in `~/.config/amp/settings.json`).
2. Start the Amp TUI (`amp`); it opens your browser to sign in to border.bot automatically.
3. To use border.bot from every Amp client, including orbs, store it on ampcode.com instead: `amp mcp remote --personal add borderbot https://api.border.bot/mcp --auth oauth`, then `amp mcp remote --personal login borderbot`.

Command:

```bash
amp mcp add borderbot https://api.border.bot/mcp
```

Config file `~/.config/amp/settings.json`:

```json
{
  "amp.mcpServers": {
    "borderbot": {
      "url": "https://api.border.bot/mcp"
    }
  }
}
```

Official setup guide: https://ampcode.com/docs/customize/mcp

Notes: Local sign-in relies on Dynamic Client Registration. Browser OAuth isn’t available inside orbs, so use the remote definition there.

### goose

By Agentic AI Foundation.

1. Click “Add to goose”, or run `goose configure` › “Add Extension” › “Remote Extension (Streamable HTTP)” and enter `borderbot` and `https://api.border.bot/mcp`.
2. In goose Desktop you can also use the sidebar › “Extensions” › “Add custom extension”.
3. Sign in to border.bot in your browser when goose asks; it registers itself automatically (Client ID Metadata Document or Dynamic Client Registration).

Config file `~/.config/goose/config.yaml`:

```yaml
extensions:
  borderbot:
    name: borderbot
    type: streamable_http
    uri: "https://api.border.bot/mcp"
    enabled: true
    timeout: 300
```

One-click install: [Add to goose](goose://extension?url=https%3A%2F%2Fapi.border.bot%2Fmcp&type=streamable_http&id=borderbot&name=borderbot&description=HS%20codes%2C%20duties%20and%20landed%20cost&timeout=300)

Official setup guide: https://goose-docs.ai/docs/getting-started/using-extensions

### Qwen Code

By Alibaba Cloud.

1. Run `qwen mcp add --scope user --transport http borderbot https://api.border.bot/mcp`.
2. Start `qwen`, enter `/mcp` and authenticate `borderbot`; sign in to border.bot in your browser.
3. Ask the model to use border.bot (restart Qwen Code first if it was already running).

Command:

```bash
qwen mcp add --scope user --transport http borderbot https://api.border.bot/mcp
```

Config file `~/.qwen/settings.json`:

```json
{
  "mcpServers": {
    "borderbot": {
      "httpUrl": "https://api.border.bot/mcp"
    }
  }
}
```

Official setup guide: https://qwenlm.github.io/qwen-code-docs/en/users/features/mcp/

Notes: OAuth endpoints are discovered and the client registers dynamically. The callback is `http://localhost:7777/oauth/callback`, so sign-in needs a local browser.

### Kimi Code CLI

By Moonshot AI.

1. Run `kimi mcp add --transport http --auth oauth borderbot https://api.border.bot/mcp`.
2. Run `kimi mcp auth borderbot` and complete the OAuth flow in your browser.
3. Run `kimi mcp test borderbot` (or `/mcp` inside Kimi Code CLI) to check the connection.

Command:

```bash
kimi mcp add --transport http --auth oauth borderbot https://api.border.bot/mcp
```

Official setup guide: https://moonshotai.github.io/kimi-cli/en/customization/mcp.html

Notes: Servers are saved to `~/.kimi/mcp.json`; OAuth tokens are kept in `~/.kimi/mcp-oauth/`.

## Automation

Agent and workflow builders that call border.bot from a step in a workflow, and an API client for trying its tools by hand.

### Microsoft Copilot Studio

By Microsoft.

1. Open your agent’s “Tools” page and select “Add a tool” › “New tool” › “Model Context Protocol”.
2. Server name: border.bot; Server description: what border.bot does (e.g. “Classifies products to HS codes and calculates duties, taxes and landed cost”); Server URL: `https://api.border.bot/mcp`.
3. Authentication: “OAuth 2.0”, type “Dynamic discovery”; select “Create”, then “Next”.
4. In “Add tool”, select “Create a new connection”, sign in to border.bot, then “Add to agent”.

Official setup guide: https://learn.microsoft.com/en-us/microsoft-copilot-studio/mcp-add-existing-server-to-agent

Notes: Copilot Studio supports the Streamable HTTP transport only. The orchestrator uses the server description to decide when to call border.bot.

### n8n

1. Add an “MCP Client Tool” to your AI Agent (or an “MCP Client” node to run one tool as a workflow step).
2. Set “Endpoint” (“MCP Endpoint URL” on the MCP Client node) to `https://api.border.bot/mcp` and “Server Transport” to “HTTP Streamable”.
3. Set “Authentication” to “MCP OAuth2”, create a new “MCP OAuth2 API” credential with “Use Dynamic Client Registration” on, and click “Connect my account” to sign in to border.bot.
4. Choose which border.bot tools to expose under “Tools to Include”.

Official setup guide: https://docs.n8n.io/integrations/builtin/cluster-nodes/sub-nodes/n8n-nodes-langchain.toolmcp/

Notes: Labels match the current node versions; n8n’s MCP Client Tool page still shows the older “SSE Endpoint” field.

### Zapier

MCP Client is in beta.

1. Go to the Apps page, click “+ Add connection”, search for and select “MCP Client”, then click “Add connection”.
2. Server URL: `https://api.border.bot/mcp`; Transport: “Streamable HTTP”; OAuth: “Yes”.
3. Click “Yes, Continue to MCP Client”, sign in to border.bot and grant Zapier access.
4. Rename the connection to border.bot, then use its tools in Zaps or Agents.

Official setup guide: https://help.zapier.com/hc/en-us/articles/38777069364109-Connect-remote-MCP-servers-to-Zapier-using-MCP-Client

Notes: MCP Client supports tool calls only.

### Make

MCP Client is in open beta for all users.

1. Add the “MCP Client” module to your scenario and click “Create a connection”.
2. Connection name: border.bot; in “MCP Server” click “+ New MCP Server” and enter `https://api.border.bot/mcp` in the “URL” field (leave “API Key / Access token” empty).
3. Click “Save”, then sign in to border.bot and confirm access.

Official setup guide: https://apps.make.com/mcp-client

Notes: Make’s OAuth redirect URI is `https://www.make.com/oauth/cb/mcp`.

### Postman

1. In a workspace, click + › “MCP” in the left sidebar, choose “HTTP” and enter `https://api.border.bot/mcp`.
2. Click “Connect” / “Load Capabilities”; for an OAuth server Postman opens the “OAuth Debugger”, click “Continue” through each phase and sign in when the “Authorization request” step opens your browser.
3. Open the “Tools” tab, pick a tool, fill in its arguments and click “Run”.

Official setup guide: https://learning.postman.com/docs/use/send-requests/protocols/mcp-requests/create/

Notes: Useful for testing border.bot’s tools outside an AI client; Postman runs the full MCP OAuth flow (protected-resource metadata, Dynamic Client Registration, PKCE).

## What your assistant can do

### Classify products

From a description or a product URL, to the code for any destination, with descriptions, confidence and alternatives.

### Calculate landed cost

Duties, taxes and fees for a code, origin and destination, with the total landed cost.

### Read product pages

Extract the title, materials, price and country of origin from a product link before classifying.

## Things to ask

- “What’s the HTS code for an insulated stainless steel water bottle made in China?”
- “Classify this product for import into the UK: https://shop.example/products/linen-shirt”
- “Calculate the landed cost of 20 units of 6109.10 from Vietnam to Germany at €12 each, with €40 shipping.”
- “Which of these product descriptions need more detail before we can classify them?”

## Billing and security

- You sign in on border.bot. The assistant never sees your password.
- Each connection is tied to one workspace that you choose when you approve it.
- Calls use the same credits and prices as the dashboard and the API, and failed calls are refunded.
- Every result is saved to the workspace history, and you can revoke a connection at any time from the dashboard.

## Frequently asked questions

### What is MCP?

The Model Context Protocol is an open standard for connecting AI assistants to tools and data. A remote MCP server, like border.bot’s, runs on the web, so you add it with a URL and sign in with OAuth instead of installing anything.

### Do I need an API key for MCP?

No. MCP clients sign in with OAuth: you log in to border.bot in your browser and approve the connection. API keys are only for calling the REST API from your own code.

### Which workspace pays for MCP calls?

The one you choose when you approve the connection. Each call uses credits from that workspace at the same prices as the dashboard and API, and the results appear in its history.

### How do I disconnect a client?

Open the dashboard to see every client connected to your workspace and revoke any of them. Revoking takes effect immediately; the client has to ask you to sign in again.

### Does my AI client’s plan support custom connectors?

Support for remote MCP servers depends on the client and sometimes on your plan or edition. Each guide on this page lists the plans its vendor documents. If the option is missing, check the vendor’s setup guide linked from that client’s steps.