The Agent Skills Directory

[COMMAND_EXECUTION]: The Python script executes the scholar CLI and a local binary (consensus) to retrieve search results. These operations are performed using secure subprocess calls with list-based arguments, preventing shell injection vulnerabilities.
[EXTERNAL_DOWNLOADS]: The script performs network requests to api.crossref.org to resolve journal names. This targets a well-known academic metadata service and is consistent with the skill's stated purpose.
[SAFE]: The skill attempts to read a local .bib file and a domain knowledge markdown file from the user's home directory and sibling skill folders. These actions are used to provide personalized search results and journal filtering, with no signs of unauthorized data exfiltration.

research