ai-index
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted text input for the purpose of linguistic analysis. This attack surface is mitigated by explicit injection guards provided in the judgment layer documentation (
references/jev-questions.md), which instruct the model to treat analyzed text as content rather than instructions to follow. - [EXTERNAL_DOWNLOADS]: The documentation includes commands for the user to fetch articles from Zenn.dev, a well-known developer platform, to establish a local analysis baseline. These references are transparent and target reputable services.
- [SAFE]: Technical audit of all included scripts (
scripts/slopscore.py,examples/rehydrate.py, and aggregation scripts) confirms they use only Python standard libraries for statistical analysis and data formatting. No obfuscation, hardcoded credentials, or malicious persistence mechanisms were found.
Audit Metadata