cut-planner
Pass
Audited by Gen Agent Trust Hub on May 11, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads and processes external data (visual transcripts and clip summaries) without explicit boundary markers or sanitization.\n
- Ingestion points: Processes multiple files in
libraries/[library-name]/summaries/summary_*.mdandlibrary.yaml.\n - Boundary markers: The instructions do not define delimiters or specific 'ignore' instructions when reading the content of summaries.\n
- Capability inventory: The skill has the ability to read and write files within the project library and can trigger the execution of the
roughcutskill.\n - Sanitization: No sanitization or validation of the text content from summaries is mentioned before it is used to update other project files.
Audit Metadata