The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user or agent to install the defuddle package globally via npm install -g defuddle. This package does not originate from a recognized trusted vendor or well-known service, making it an unverified dependency.
[COMMAND_EXECUTION]: The instructions involve executing the defuddle command-line utility, which includes passing user-supplied URLs as arguments and supports writing output to local files (e.g., -o content.md).
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to fetch and parse content from external web pages, creating a significant surface for indirect prompt injection attacks.
Ingestion points: Content is ingested from arbitrary external URLs via the defuddle parse command.
Boundary markers: There are no specific instructions or delimiters provided to the agent to help distinguish between valid content and potential embedded instructions in the fetched data.
Capability inventory: The skill provides access to CLI execution and local file writing capabilities, which could be abused if malicious instructions are processed.
Sanitization: The skill lacks explicit instructions for sanitizing, escaping, or validating the extracted markdown content before it is processed by the agent.

defuddle