The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted news data from web searches without explicit boundary markers. * Ingestion points: External news signals and sources gathered during the research phase. * Boundary markers: No delimiters or explicit instructions to ignore embedded commands were identified. * Capability inventory: The skill has the ability to write files to the local system (mkdir, cat) and perform network operations via web search. * Sanitization: No specific sanitization or filtering logic for external content is defined.
[COMMAND_EXECUTION]: The skill executes shell commands like mkdir and cat to manage a local output directory and save generated report files. These commands use subshell execution for dynamic timestamp generation.
[EXTERNAL_DOWNLOADS]: The skill utilizes web search tools to retrieve news articles and trend information from the internet. This behavior is a core component of its research functionality.

coolhunter