sequence-load
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources, specifically prospect-refresh HTML reports and Gmail drafts, to extract lead metadata and perform automated outreach actions.
- Ingestion points: Stage 1 reads prospect-refresh HTML reports and Gmail drafts to identify name, email, and company details.
- Boundary markers: No explicit delimiters or instructions to ignore embedded prompts are present in the processing logic to distinguish data from instructions.
- Capability inventory: The skill utilizes tools for HubSpot CRM management and Apollo sales automation (contact creation and sequence enrollment). It also performs local file writes to
~/.claude/skill-analytics/for outcome reporting. - Sanitization: While the skill validates structured fields like email and phone formats, it does not implement sanitization or filtering for natural language fields extracted from the external reports that could contain malicious instructions.
Audit Metadata