research-start
Pass
Audited by Gen Agent Trust Hub on Jul 3, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits a vulnerability surface for indirect prompt injection by design.
- Ingestion points: Step 4 of the workflow instructs the agent to read and synthesize "existing research uploaded" by the student.
- Boundary markers: Absent. The instructions do not specify the use of delimiters or clear separation between the agent's instructions and the untrusted student data.
- Capability inventory: Across the skill, there are no detections of subprocess execution, network operations (curl/wget), or file writing capabilities. The agent is limited to reading local files and generating text responses.
- Sanitization: Absent. There are no mechanisms defined to escape or filter potentially malicious instructions embedded within the student-uploaded research documents.
Audit Metadata