phishing-triage
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted text from suspicious emails, SMS, and other messages, which creates a surface for potential indirect prompt injection attacks.
- Ingestion points: The skill explicitly requests the content of suspicious messages in the Required Inputs section of SKILL.md.
- Boundary markers: The skill does not utilize formal data delimiters (such as XML tags or markdown blocks) to isolate untrusted input, though it instructs the agent to treat the input as data for analysis.
- Capability inventory: The skill possesses no capabilities for command execution, file system modification, or network requests, effectively neutralizing the impact of potential injections.
- Sanitization: No specific sanitization or filtering logic is applied to the message content before processing.
- [SAFE]: The skill incorporates security best practices by explicitly instructing the agent never to recommend clicking suspicious links and to always suggest verification through independent, official channels.
Audit Metadata