skills/silvabyte/skills/handoff/Gen Agent Trust Hub

handoff

Pass

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes conversation history which is considered untrusted input. While it does not use explicit boundary markers to delimit this data, it incorporates safety instructions to redact sensitive information such as API keys and PII.
  • Ingestion points: Current conversation history.
  • Boundary markers: None specified in the instructions.
  • Capability inventory: File system write access (saving to the OS temporary directory).
  • Sanitization: Instructions to redact sensitive information (API keys, passwords, PII) are present.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 3, 2026, 04:19 PM
Security Audit — agent-trust-hub — handoff