escalation-flagger
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external content from contract clauses and review memos to identify routing needs. While this is an ingestion surface for untrusted data, the skill is not granted any dangerous tools (such as network, shell, or file write operations) that could be exploited by embedded instructions. Furthermore, the instructions mandate that the agent only draft messages for human review, preventing automated execution of potentially malicious content.
- [SAFE]: The skill logic is declarative and restricted to text analysis and drafting. No signs of obfuscation, credential exposure, persistence mechanisms, or unauthorized privilege escalation were found.
Audit Metadata