The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install the defuddle package globally using npm install -g defuddle.
[COMMAND_EXECUTION]: The skill operates by executing the defuddle CLI tool through the shell to parse URLs and save output to files.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external, untrusted web pages, which is a common attack surface for indirect prompt injection.
Ingestion points: External URLs processed via defuddle parse <url> (SKILL.md).
Boundary markers: Absent. There are no instructions for the agent to use delimiters or safety markers when processing the extracted text.
Capability inventory: The skill has shell execution capabilities to run the defuddle command.
Sanitization: While the tool removes HTML clutter like ads and navigation, it does not sanitize the remaining natural language content for malicious instructions directed at the LLM.

defuddle