equivariant-architecture-designer
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references several specialized machine learning libraries such as
e3nn,escnn,torch-geometric,NequIP,SchNet, andMACE. These are well-known and reputable open-source projects in the equivariant neural network community and are considered safe sources. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest untrusted data in the form of user-provided symmetry group specifications and task requirements (identified in Step 1 of SKILL.md). However, an analysis of the skill's capabilities reveals no access to dangerous tools such as network operations, file writing, or shell execution. Consequently, there is no viable attack surface for indirect prompt injection to exploit.
- [SAFE]: No evidence of malicious intent, credential theft, obfuscation, or persistence mechanisms was detected. The instructions and code templates follow standard, best-practice development patterns for the intended domain.
Audit Metadata