subtitle-extractslator
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches its primary runtime components, such as the SubtitleExtractslator CLI and the Loom SO orchestrator, from the author's official GitHub repositories (github.com/waynebaby/SubtitleExtractslator and github.com/waynebaby/Techne-Loom).
- [COMMAND_EXECUTION]: The agent is instructed to use the dotnet runtime to execute external binaries and workflows for media probing, extraction, and translation tasks.
- [INDIRECT_PROMPT_INJECTION]: The skill processes subtitle content retrieved from OpenSubtitles and embedded media tracks, which are untrusted external inputs.
- Ingestion points: External subtitle data is ingested via OpenSubtitles API downloads and media file probing as specified in references/mcp.md.
- Boundary markers: Employs structural validation and timing checks (references/troubleshooting.md) to verify subtitle integrity before processing.
- Capability inventory: The skill maintains capabilities for filesystem access (reading media/writing SRT) and command execution (dotnet CLI).
- Sanitization: Uses deterministic parsing to ensure cue ordering and timing rhythm are preserved, reducing the impact of malformed data.
- [DYNAMIC_EXECUTION]: Orchestration logic is driven by a JSON template (so-template.json) interpreted at runtime by the Loom SO engine, allowing for complex conditional branching and state management.
Audit Metadata