screen-recording-analysis
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted text data extracted from screen recordings (OCR text) which presents a potential surface for indirect prompt injection.
- Ingestion points: The skill ingests data extracted from video stills and transcripts, as described in the 'Private data' section of
SKILL.md. - Boundary markers: The instructions do not mandate the use of specific boundary markers or delimiters when the agent writes extracted text to intermediate files for processing.
- Capability inventory: The skill utilizes local Python scripts (
actions.py,private_text.py) which are executed viapython3to analyze the processed data. - Sanitization: The skill provides a robust PII detection script (
private_text.py) that uses regular expressions and the Luhn algorithm to find and mask sensitive information such as API keys, emails, and credit card numbers before the agent further interacts with the data.
Audit Metadata