duoshe
Pass
Audited by Gen Agent Trust Hub on May 15, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill is designed to ingest and process data from external, untrusted scholarly sources such as research PDFs, abstracts, and web links. This represents a potential surface for indirect prompt injection where adversarial text within a paper could attempt to influence the agent's output.
- Ingestion points: SKILL.md (processes Google Scholar queries, URLs, PDF paths, and abstracts).
- Boundary markers: The skill includes instructions to strictly separate verified claims from inference and to label abstract-only data.
- Capability inventory: The skill performs network operations (fetching academic metadata) and local filesystem operations (creating folders and writing Markdown files).
- Sanitization: No specific technical sanitization of external content is defined beyond instructional grounding.
- [COMMAND_EXECUTION]: Local Filesystem Operations. The skill automates the creation of a local directory structure and the generation of multiple Markdown files to store research analysis. This behavior is documented as the primary workflow of the skill and is intended for user productivity.
Audit Metadata