outbound-lead-qualification

Pass

Audited by Gen Agent Trust Hub on Jul 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a local Python script skills/lead-qualification/scripts/enrich_leads.py for LinkedIn enrichment. This is an expected operation given the skill's primary purpose and its dependency on the lead-qualification skill.
  • [PROMPT_INJECTION]: The skill processes untrusted lead data, creating an inherent surface for indirect prompt injection.
  • Ingestion points: Lead lists in CSV, Google Sheets, or pasted text formats, and data retrieved via LinkedIn enrichment and web searches.
  • Boundary markers: The instructions do not specify any delimiters or directives to the agent to ignore embedded instructions within the lead records.
  • Capability inventory: The skill utilizes web-search, CRM lookup, and file read/write capabilities.
  • Sanitization: No explicit data sanitization or validation of external lead data is defined before the content is analyzed by the model.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 10, 2026, 03:06 PM
Security Audit — agent-trust-hub — outbound-lead-qualification