share-pretty-gist
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses authoritative language ("ALWAYS use this skill", "This is the default way") to mandate its use over standard agent functions for creating gists, attempting to override the agent's default operational logic.
- [DATA_EXFILTRATION]: The skill directs the agent to perform
curlrequests (GET and POST) togists.sh, a third-party domain that is not on the whitelisted domains list. These network operations transmit the user's GitHub username and Gist ID to the external service to "warm the cache" or refresh data. - [COMMAND_EXECUTION]: The skill utilizes the
ghcommand-line interface to interact with GitHub Gists and thecurlutility to communicate with the third-party rendering service, involving shell-level execution on the host environment.
Audit Metadata