The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill directs the agent to install a third-party Node.js package (defuddle) using npm install -g (SKILL.md). This package is not managed by a verified trusted vendor, posing a supply-chain risk as the installed binary executes with the user's local permissions.
[COMMAND_EXECUTION]: The skill instructions involve executing the defuddle CLI tool (SKILL.md) to parse URLs provided by the user. This creates a dependency on an external binary whose security and behavior are not locally verified.
[PROMPT_INJECTION]: The skill's primary function is to process data from arbitrary external URLs, creating a surface for indirect prompt injection.
Ingestion points: Content is retrieved from external URLs using the defuddle parse command (SKILL.md).
Boundary markers: None are provided to help the agent distinguish between the parsed web content and its system instructions.
Capability inventory: The skill enables the agent to execute shell commands to interact with external network resources.
Sanitization: The retrieved markdown content is not sanitized or filtered, which could allow malicious instructions embedded on a web page to hijack the agent's logic.

defuddle