inbound-lead-audit-cx-map
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is designed to ingest data from external sources, specifically CRM timelines, routing evidence, and tracker rows as specified in
SKILL.mdandmetadata.yaml. This creates a potential surface for indirect prompt injection; however, the skill significantly mitigates this risk by explicitly instructing the agent to keep lead data private and mandating human approval before any CRM updates or message sending. - [EXTERNAL_DOWNLOADS]: The
SETUP.mdfile provides instructions to utilize the Zapier SDK vianpx zapier-sdk login. This is a legitimate tool provided by the vendor for managing developer workflows. Additionally, the skill utilizes a Zapier MCP server hosted athttps://mcp.zapier.com, which is an official vendor domain. - [SAFE]: The workflow design adheres to the principle of least privilege by recommending read-only tests and draft-only work before enabling writes. The instructions in
README.mdandSKILL.mdprioritize human-in-the-loop governance for all sensitive operations.
Audit Metadata