publish
Fail
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill instructs the agent to install a utility by piping a remote script ('https://treg.to/install.sh') directly to the shell ('curl -fsSL ... | sh'). This 'curl pipe sh' pattern is a significant security risk as it executes unverified code from an external source.
- [EXTERNAL_DOWNLOADS]: In addition to the installation script, the skill downloads an MCP configuration from 'https://treg.to/mcp/' and connects to external APIs for Instagram, TikTok, and YouTube.
- [COMMAND_EXECUTION]: The skill requires extensive use of shell commands, including 'ffmpeg' and 'ffprobe' for media processing, 'treg' for account management and publishing, and system commands like 'open' or 'explorer' to interact with the local file system.
- [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted external data, creating an injection surface.
- Ingestion points: Reads video transcripts, scripts, and the 'post.md' file in Step 4.
- Boundary markers: The instructions lack delimiters or warnings to ignore instructions embedded within the processed text files.
- Capability inventory: The skill can write files ('social.mp4', 'cover.jpg', 'post.md', 'published.md'), execute media tools ('ffmpeg'), and perform network operations via the 'treg' CLI.
- Sanitization: No sanitization or escaping of the ingested text is performed before it is used to generate captions or plans.
Recommendations
- HIGH: Downloads and executes remote code from: https://treg.to/install.sh - DO NOT USE without thorough review
- AI detected serious security threats
Audit Metadata