cross-ref-topic-ledger

Pass

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill defines a logical workflow for data processing and tagging within the agent's environment. It does not perform network operations, access sensitive credentials, or execute code.
  • [INDIRECT_PROMPT_INJECTION]: The workflow involves ingesting external candidate summaries and a local ledger file. The instructions do not define explicit delimiters or boundary markers to isolate this external content from the agent's instructions, though the skill's read-only scope limits the potential for exploitation.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 3, 2026, 02:36 PM
Security Audit — agent-trust-hub — cross-ref-topic-ledger