The Agent Skills Directory

[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by design, as it fetches and processes content from external, untrusted web pages and injects that content into the agent's context.
Ingestion points: The defuddle parse <url> command retrieves raw data from any URL provided by the user.
Boundary markers: The instructions do not specify any delimiters (e.g., XML tags or clear markdown blocks) to isolate the fetched external content from the agent's core instructions.
Capability inventory: The skill uses shell command execution via the defuddle CLI.
Sanitization: No sanitization or filtering logic is mentioned to prevent malicious instructions embedded in web pages from influencing the agent's behavior.
[COMMAND_EXECUTION]: The skill uses shell commands to interact with the defuddle utility. Specifically, it uses defuddle parse <url> where the URL is a user-controlled parameter. This could lead to command injection if the execution environment does not properly escape the URL string.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of an external dependency, the defuddle package, via a global NPM install (npm install -g defuddle). This introduces a dependency on a package in the public registry.

defuddle