caption-polish
Pass
Audited by Gen Agent Trust Hub on Jun 20, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill's logic is confined to editorial cleanup of local subtitle and transcript files. It lacks capabilities for code execution, system modification, or network communication.\n- [PROMPT_INJECTION]: The skill ingests untrusted data from 'audio/captions.json' and 'audio/voice.json', creating a surface for indirect prompt injection. However, the risk is negligible as the agent uses this data solely for string manipulation and JSON formatting.\n
- Ingestion points: The skill reads external data from 'audio/captions.json', 'audio/voice.json', and 'script.md'.\n
- Boundary markers: No explicit delimiters or 'ignore embedded instructions' warnings are provided for the input data.\n
- Capability inventory: File operations are restricted to reading and writing local JSON and Markdown files. No subprocess calls, remote code execution, or network operations are present.\n
- Sanitization: The skill implements specific pattern matching and formatting rules for artifact removal and spelling correction, rather than general content filtering.
Audit Metadata