plotloom-subtitle-burnin
Pass
Audited by Gen Agent Trust Hub on Jun 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell commands to run
ffmpegandstable-tsfor video processing tasks. It employs variables like$VIDEOand$OUTin shell strings, which requires proper path escaping by the agent to ensure safe execution. - [DATA_EXFILTRATION]: The skill proactively instructs the agent to redact sensitive information such as API keys, file tokens, and OAuth tokens from any delivery messages to prevent accidental data exposure.
- [PROMPT_INJECTION]: The skill processes external files such as scripts and subtitle files (SRT, ASS) for timing and text correction, which creates a potential surface for indirect prompt injection if the files contain malicious instructions.
- Ingestion points:
video-prompts-en.md, SRT, and ASS files. - Boundary markers: No delimiters or specific ignore-instruction warnings are specified for the processed file content.
- Capability inventory: Execution of system commands through subprocesses.
- Sanitization: No specific filtering or escaping of input file content is defined.
Audit Metadata