hackathon-idea-scoring
Pass
Audited by Gen Agent Trust Hub on Mar 12, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions define a purely logical process for ranking and scoring ideas based on user-provided inputs.
- [SAFE]: No indicators of data exfiltration, credential theft, or remote code execution were found.
- [PROMPT_INJECTION]: The skill processes untrusted data which creates a surface for indirect prompt injection. 1. Ingestion points: 'ideas' and 'evaluation_axes' inputs in SKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: No dangerous capabilities (subprocess, network, file-write) identified across the skill. 4. Sanitization: Absent. The lack of capabilities results in a safe assessment.
Audit Metadata