video-frames

Pass

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses a bash script (scripts/frame.sh) to execute the ffmpeg binary. While the script uses double quotes to wrap variables and set -euo pipefail to prevent some shell errors, it interpolates the $index variable directly into a complex filter string (-vf "select=eq(n\\,${index})"), which could allow for ffmpeg filtergraph injection if the input is not strictly numeric.
  • [INDIRECT_PROMPT_INJECTION]:
  • Ingestion points: The script scripts/frame.sh processes external video files and user-controlled parameters including --time, --index, and --out.
  • Boundary markers: Absent. The skill does not instruct the agent to validate the structure or safety of the provided paths or parameters.
  • Capability inventory: The skill has the capability to execute shell commands (ffmpeg) and write files to the system via mkdir -p and ffmpeg's output redirection.
  • Sanitization: Partial. The script uses standard shell quoting for file paths, but does not perform validation to ensure timestamps follow the HH:MM:SS format or that indices are strictly integers.
  • [EXTERNAL_DOWNLOADS]: The skill metadata (SKILL.md) references the official ffmpeg.org homepage and provides instructions to install the dependency via the Homebrew package manager (brew install ffmpeg).
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 19, 2026, 11:31 PM
Security Audit — agent-trust-hub — video-frames