container-app-lightpanda

A Privasys container app that wraps the Lightpanda headless browser as an AI tool (MCP server).

Overview

This container exposes the Lightpanda headless browser as MCP-compatible AI tools that can be deployed to Enclave OS Virtual (TDX) enclaves. AI agents can discover and invoke these tools through the Privasys platform with full hardware attestation.

Tools

Tool	Description	Endpoint
browse	Fetch a web page and return its content as markdown or HTML	POST /browse

browse

Fetches a web page using Lightpanda's headless browser engine and returns the content in the requested format.

Request:

{
  "url": "https://example.com",
  "format": "markdown"
}

Response:

{
  "url": "https://example.com",
  "format": "markdown",
  "content": "# Example Domain\n\nThis domain is for use in illustrative examples..."
}

Parameters:

• url (string, required) — The URL to fetch. Must start with http:// or https://.
• format (string, optional) — Output format: "markdown" (default) or "html".

Container MCP Manifest

The privasys.json file declares the available AI tools following the Privasys container MCP standard. This manifest is provided when creating the app on the platform and enables:

• AI Tools (MCP) — AI agents discover and invoke tools via the MCP protocol
• API Testing — Developers can test tools directly from the developer portal

Building

docker build -t container-app-lightpanda .

Running Locally

docker run -p 8080:8080 container-app-lightpanda

Test the browse tool:

curl -X POST http://localhost:8080/browse \
  -H "Content-Type: application/json" \
  -d '{"url": "https://example.com", "format": "markdown"}'

Deploying to Privasys

1. Push the Docker image to a registry
2. Create a container app on the platform with the privasys.json manifest
3. Deploy to a TDX enclave

License

AGPL-3.0 — See LICENSE for details.