shot-list
Pass
Audited by Gen Agent Trust Hub on Feb 21, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- PROMPT_INJECTION (LOW): Indirect prompt injection surface identified through the ingestion and parsing of screenplay files.
- Ingestion points: The
parse_screenplayfunction in 'SKILL.md' extracts content from user-provided files (.fountain, .fdx, .txt, .pdf, .docx). - Boundary markers: Absent; the skill does not wrap parsed script content in protective delimiters or provide instructions to the model to ignore embedded commands.
- Capability inventory: The skill possesses file-writing capabilities (PDF and CSV export) via 'scripts/generate_shot_list_pdf.py'.
- Sanitization: Absent; raw text is extracted via regex and used directly in the collaborative planning phase.
- SAFE (SAFE): No evidence of malicious behavior was found. There are no external network connections, credential exposures, obfuscated code, or unauthorized persistence mechanisms. All logic is transparent and aligned with the stated purpose of shot list generation.
Audit Metadata