make-video

Pass

Audited by Gen Agent Trust Hub on Nov 19, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads content from external URLs for two primary purposes: checking for software updates and fetching media assets for video generation.
  • Evidence: scripts/update-check fetches version information from the author's repository at https://raw.githubusercontent.com/Atmanity/atmee-toolkit/main/plugins/atmee/VERSION.
  • Evidence: scripts/media.sh utilizes curl to download user-specified images or audio files, implementing size limits (60MB) and protocol restrictions (HTTPS only) to enhance security.
  • [REMOTE_CODE_EXECUTION]: The update utility suggests a command for installing or updating the Atmee CLI via a piped shell script from the vendor's GitHub repository.
  • Evidence: scripts/update-check provides the command curl -fsSL https://raw.githubusercontent.com/Atmanity/atmee-toolkit/main/install.sh | sh. This is a vendor-owned resource used for legitimate maintenance and is only suggested for execution following user confirmation.
  • [COMMAND_EXECUTION]: The skill relies on local Bash scripts to manage files, detect the agent's host environment, and perform authenticated uploads.
  • Evidence: scripts/media.sh and scripts/upload-file.sh execute various shell commands (curl, mkdir, mv, wc) with strict input validation, including regex checks on filenames and verification that file operations remain within the project's directory structure.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it processes untrusted user-provided media files and prompts for video generation.
  • Ingestion points: User-provided images and recordings ingested via upload-file.sh or through URLs processed by media.sh.
  • Boundary markers: The skill instructs the agent to draft prompts and scripts with the user, but explicit boundary markers for the generated content are absent in the prompt templates.
  • Capability inventory: The skill has the ability to execute shell scripts, perform network operations, and write files within the workspace.
  • Sanitization: Effective sanitization is present in the supporting scripts, which prevent symlink attacks, directory traversal, and forbidden filename patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 19, 2026, 10:58 PM