compose-creator-layer
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/compose.pyusessubprocess.runto execute theffmpegandffprobebinaries for video editing tasks. The arguments are passed as a list of strings, which is a secure implementation that prevents shell injection vulnerabilities.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted JSON data from a user-provided file to determine timing and layout for video compositing.\n - Ingestion points: The script reads and parses data from the JSON file specified by the
--beatsargument and metadata from the file associated with--layerinscripts/compose.py.\n - Boundary markers: None identified; the skill assumes the provided files follow a valid JSON structure.\n
- Capability inventory: The skill has the ability to execute
ffmpegandffprobesubprocesses and write video files to the disk based on the configuration.\n - Sanitization: Values extracted from the JSON input are treated as typed data (converted to integers or floats) and interpolated into
ffmpegfilter strings using strict format specifiers (e.g.,%d,%.3f), effectively neutralizing them as potential code injection vectors.
Audit Metadata