sequence-performance
Pass
Audited by Gen Agent Trust Hub on Jun 18, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it is designed to ingest and analyze untrusted content from external email recipients.\n
- Ingestion points: In Step 1 (Pull Campaign Data) and Step 3 (Reply Analysis), the skill fetches and reads the
reply_textfrom external outreach tools like Smartlead and Instantly.\n - Boundary markers: The instructions in Step 3 for 'Reply Classification' and 'Objection Pattern Analysis' do not define boundary markers or instruct the agent to ignore potentially malicious instructions embedded within the reply text.\n
- Capability inventory: The skill produces reports and passes data to other skills such as
email-drafting, creating a chain where a malicious reply could influence downstream drafting or outreach actions.\n - Sanitization: No explicit sanitization, filtering, or validation of the retrieved reply content is mentioned in the sequence of operations.
Audit Metadata