sillytavern-card-components
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill defines a robust security posture by mandating a "safe disassembly" process. It requires the use of fresh sandboxes and explicit output roots to prevent accidental or malicious file system writes to active source trees or shared libraries.
- [SAFE]: Provenance and integrity are maintained through the requirement of evidence snapshots, which include recording the SHA-256 hash and byte size of input artifacts before any processing occurs.
- [SAFE]: The instructions explicitly address path traversal risks by rejecting absolute paths and parent traversal patterns during component dependency resolution and artifact emission.
- [SAFE]: The skill recognizes the risks of processing untrusted data (SillyTavern cards) and implements mitigations such as "modelVisible" flags in the registry to control whether processed content can enter the AI's model context, effectively limiting the surface for indirect prompt injection.
- [SAFE]: The skill defines a clear runtime dependency contract that isolates embedded scripts (regex, helper scripts) from the host environment, preventing unauthorized capability elevation during card management.
Audit Metadata