instagram-gallery-download
Fail
Audited by Gen Agent Trust Hub on Jun 20, 2026
Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The
SKILL.mdfile explicitly mandates the use of escalated sandbox permissions (require_escalated). This configuration allows the agent and its associated scripts to execute commands outside of the default restricted environment. - [COMMAND_EXECUTION]: The helper script
scripts/download_instagram_gallery.shimplements an argument passthrough that forwards any input following the--delimiter directly to thegallery-dlbinary. - [COMMAND_EXECUTION]: The
gallery-dltool supports flags such as--execand--exec-afterfor executing arbitrary shell commands. Combined with the script's passthrough and the requested escalated permissions, this enables a user to perform arbitrary command execution on the host. - [PROMPT_INJECTION]: The skill represents an indirect prompt injection surface as it ingests untrusted data from Instagram URLs. Ingestion points: Instagram post metadata processed by
gallery-dl. Boundary markers: Absent. Capability inventory: Escalated shell execution and file writing. Sanitization: Absent. This surface could allow maliciously crafted metadata to influence agent behavior.
Recommendations
- AI detected serious security threats
Audit Metadata