curator-skill
Pass
Audited by Gen Agent Trust Hub on Apr 11, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides a mechanism to update its persona database by fetching a JSON file from the author's official repository (github.com/Aar0nPB/curator-skill). This download is triggered only by explicit user request and fetches data used to update the recommendation catalog.
- [COMMAND_EXECUTION]: The skill generates and presents shell commands (e.g.,
npx skills add [repo] -g -y) to the user. These commands are intended to allow the user to install other persona skills listed in the catalog. The commands are shown to the user for manual confirmation rather than being executed silently in the background. - [DATA_INJECTION_SURFACE]: The skill processes a catalog of persona descriptions, triggers, and installation commands stored in
persona-recommendations.json. Because this data can be updated from a remote source, it serves as a surface where external content influences the agent's output. However, since the source is the author's verified repository, the risk is minimized. - [SAFE]: The identified network and command activities are functional components of the skill's primary use case. The instructions in
SKILL.mdexplicitly direct the agent to avoid automated network calls during routine operations, prioritizing local files to ensure performance and transparency.
Audit Metadata