sdr-email-sms-playbook-skill
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface where untrusted data from external sources (prospect enrichment, organization names, and personalization hooks) is interpolated into prompts to generate content for sensitive tools.
- Ingestion points: The skill ingests data from the
qualify_leadtool and user-provided personalization hooks (triggers like budget cycles or failed recordings) as described in the Quick Start and Touch 1 template sections. - Boundary markers: There are no explicit instructions to use XML tags, markdown blocks, or other delimiters to isolate untrusted prospect data from the core system instructions within the prompt interpolation logic.
- Capability inventory: The skill utilizes
mcp__claude_ai_Gmail__create_draftto create drafts in the user's Gmail account, representing a high-integrity capability that could be influenced by malicious input in the lead data. - Sanitization: The instructions do not define any sanitization, escaping, or validation protocols for the external content before it is processed by the LLM to generate email bodies.
Audit Metadata