gaokao-science-tutor
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process student queries, creating a surface for potential indirect prompt injection. However, since the skill has no access to tools, network operations, or file-writing capabilities, the impact is restricted to persona disruption. Ingestion points: Student questions provided via dialogue in SKILL.md. Boundary markers: None; user input is treated as conversation within the prompt template. Capability inventory: None; the skill lacks tool definitions and does not perform network or file operations. Sanitization: None.
- [EXTERNAL_DOWNLOADS]: The README documentation suggests cloning the official GitHub repository (github.com/flysheep-ai/education-skills.git) for installation purposes. This resource belongs to the skill's author and is provided for legitimate setup.
Audit Metadata