paper-to-skill
Pass
Audited by Gen Agent Trust Hub on Apr 12, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: Potential for indirect prompt injection via untrusted source papers (PDFs, URLs, or text) that are processed and used to generate instructions for the test-engineer agent. * Ingestion points: External content from arXiv, URLs, or user input as defined in SKILL.md. * Boundary markers: Uses block quotes for agent hand-off instructions but lacks explicit guardrails against adversarial content in the paper source. * Capability inventory: File system writes at 'skills/{name}/' and network fetching via WebFetch. * Sanitization: Extraction into a structured YAML specification provides a layer of data isolation.- [SAFE]: Downloads and references are directed at well-known academic services like arxiv.org for research purposes.
Audit Metadata