sa-token-advanced
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill includes a self-referential evaluation report (
evaluation-report.html) that presents internal safety scores and trustworthiness ratings. Such content is noted as a potential attempt to influence automated security analyzers. - Ingestion points: Documentation files and the included evaluation report file.
- Boundary markers: None provided to separate framework examples from instructions.
- Capability inventory: The framework examples cover sensitive operations such as identity switching (
switchTo), session termination, and service-specific account banning. - Sanitization: The provided code snippets demonstrate raw parameter usage from web requests (e.g.,
userId), which requires developers to implement their own application-level validation. - [EXTERNAL_DOWNLOADS]: The skill links to project documentation and source code on GitHub and fetches styling resources from a public CDN for the included HTML report.
Audit Metadata