problem-parser
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data such as problem statements, excerpts, and screenshots to generate structured artifacts.
- Ingestion points: Problem statements, attachment descriptions, OCR text, and user-provided notes defined in the 'Inputs' section.
- Boundary markers: The skill does not employ specific delimiters or 'ignore instructions' warnings for the input data, although it uses sentinels like '[MODELER INPUT NEEDED]' in its output to require human intervention.
- Capability inventory: The agent is permitted to read and write files within the 'workspace/' directory but lacks network access, shell execution, or privilege escalation capabilities.
- Sanitization: There is no evidence of input validation, filtering, or escaping to prevent instructions embedded in the problem statements from influencing the agent's behavior.
Audit Metadata