The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed around the camoufox-cli tool for browser automation. The tool includes high-privilege commands for accessing cookies, exporting browser identities, and performing file uploads, which are necessary for the skill's functionality but increase the security surface area.- [DYNAMIC_EXECUTION]: The documentation provides instructions on using the eval command to execute JavaScript within the target web page context. It specifically teaches the agent to wrap multi-step logic in Immediately Invoked Function Expressions (IIFEs) to ensure execution, which allows for complex runtime code execution on external sites.- [INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface for indirect prompt injection by processing data from untrusted websites. 1. Ingestion points: Page content is ingested via the snapshot and text commands (SKILL.md). 2. Boundary markers: The skill does not define explicit delimiters or instructions to ignore embedded commands in the web data. 3. Capability inventory: The agent is granted capabilities to click, fill, type, upload, and execute JavaScript (eval). 4. Sanitization: No mechanisms for sanitizing or validating external content are specified before the data is processed by the agent's logic.

browser-guide