Plugins & connectors

Give your AI eyes on the web

Shotlane plugs into ChatGPT, Claude and every MCP-capable tool. Your assistant can then screenshot any page (and show it to you), read pages as clean Markdown — even JavaScript-heavy ones — and create PDFs. Connecting takes one click: sign in or create a free account when prompted.

ChatGPT

1. In ChatGPT, open Settings → Apps & Connectors (called Plugins in newer versions) and enable Developer mode under Advanced settings if your plan requires it.
2. Choose Create / Add custom connector, name it Shotlane, and paste https://shotlane.dev/mcp. Authentication: OAuth.
3. Click Connect, then sign in or create a free Shotlane account in the window that opens.
4. In a chat, enable Shotlane from the tools menu and ask: "Take a screenshot of stripe.com on mobile". Screenshots appear inline with a full-size link.

Menu names vary by ChatGPT plan and version. Custom connectors may be limited to certain plans.

Claude (claude.ai & Claude Desktop)

1. Open Settings → Connectors and choose Add custom connector.
2. Name: Shotlane · URL: https://shotlane.dev/mcp, then Add and Connect.
3. Sign in or create your free account, approve, and you're done. Try: "Read https://react.dev/learn and summarize the key ideas."

Claude Code

claude mcp add --transport http shotlane https://shotlane.dev/mcp
# then run /mcp inside Claude Code and choose "Authenticate"

Prefer an API key? claude mcp add --transport http shotlane https://shotlane.dev/mcp --header "X-API-Key: sl_…"

Cursor, Windsurf & other editors

Cursor — ~/.cursor/mcp.json (OAuth sign-in starts automatically):

{
  "mcpServers": {
    "shotlane": { "url": "https://shotlane.dev/mcp" }
  }
}

VS Code (GitHub Copilot) — .vscode/mcp.json:

{
  "servers": {
    "shotlane": { "type": "http", "url": "https://shotlane.dev/mcp" }
  }
}

Gemini CLI — ~/.gemini/settings.json:

{
  "mcpServers": {
    "shotlane": { "httpUrl": "https://shotlane.dev/mcp", "headers": { "X-API-Key": "sl_your_key" } }
  }
}

Windsurf — mcp_config.json: {"mcpServers": {"shotlane": {"serverUrl": "https://shotlane.dev/mcp"}}}

Any client that can't do OAuth can authenticate with your API key, as an X-API-Key header or Authorization: Bearer sl_….

Your own agents

OpenAI Responses API — remote MCP tool:

tools=[{
    "type": "mcp",
    "server_label": "shotlane",
    "server_url": "https://shotlane.dev/mcp",
    "headers": {"X-API-Key": os.environ["SHOTLANE_API_KEY"]},
    "require_approval": "never",
}]

LangChain (langchain-mcp-adapters):

client = MultiServerMCPClient({"shotlane": {
    "transport": "streamable_http",
    "url": "https://shotlane.dev/mcp",
    "headers": {"X-API-Key": os.environ["SHOTLANE_API_KEY"]},
}})
tools = await client.get_tools()

Claude API — use the MCP connector with url: https://shotlane.dev/mcp and your API key as the authorization token. Or skip MCP entirely and call the REST API or OpenAPI spec from any framework.

Agent Skill

A ready-made skill (SKILL.md + a dependency-free CLI script) that teaches agents when to screenshot vs. read a page, how to handle mobile layouts, slow pages and errors. Works with Claude (Skills), ChatGPT plugins that bundle skills, Codex and Claude Code.

Download shotlane.zip   View SKILL.md

Custom GPTs (Actions)

Building your own GPT? Import https://shotlane.dev/gpt/openapi.json as an Action. It uses OAuth so each user connects their own Shotlane account; email us for OAuth client credentials for your GPT.