The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it retrieves and processes content from arbitrary external websites.
Ingestion points: Data is ingested from websites through browser tools called via mcpClient.callTool in script.js.
Boundary markers: The skill does not implement boundary markers or instructions to the agent to disregard instructions found within the processed data.
Capability inventory: The skill possesses the capability to control a browser instance via CDP and execute various tools provided by the remote MCP server.
Sanitization: There is no evidence of sanitization or filtering of the retrieved website content before it is passed to the AI agent.
[DATA_EXFILTRATION]: The skill transmits the user-defined CDP_URL (which identifies the browser's control interface) to the SERVER_URL in the x-cdp-url HTTP header during the connection phase.
[EXTERNAL_DOWNLOADS]: The script connects to a remote MCP server at a dynamically provided SERVER_URL to fetch tool definitions and execute remote logic.
[COMMAND_EXECUTION]: The skill facilitates the execution of browser actions and tool-based commands through a combination of the Chrome DevTools Protocol and the Model Context Protocol.

browser-ability