The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill initiates the download and installation of a Go binary from an unverified third-party repository.
Evidence: The metadata section contains a go install module pointing to github.com/Hyaxia/blogwatcher/cmd/blogwatcher@latest. The markdown body also instructs the user to run go install github.com/Hyaxia/blogwatcher/cmd/blogwatcher@latest.
[COMMAND_EXECUTION]: The skill relies on executing the blogwatcher binary to perform its primary functions.
Evidence: Multiple command examples are provided, including blogwatcher add, blogwatcher scan, and blogwatcher articles.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted data from external RSS/Atom feeds.
Ingestion points: Content is ingested via the blogwatcher scan and blogwatcher articles commands which fetch data from user-defined feed URLs (SKILL.md).
Boundary markers: There are no explicit instructions or delimiters to prevent the agent from obeying commands embedded within the fetched feed articles.
Capability inventory: The skill has the capability to execute the blogwatcher CLI and display its output to the agent (SKILL.md).
Sanitization: No sanitization or content filtering mechanisms are described or implemented for the external feed data.

blogwatcher