job-post-builder
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection Surface: The skill's workflow involves researching external job postings via web search and searching local files in Google Drive or on the Desktop to inform its drafts. This ingestion of untrusted data creates a potential surface where instructions embedded in those external sources could influence the agent's behavior.\n
- Ingestion points: Phase 2 (web search for comparable postings and local file search for existing JDs/templates in Google Drive and Desktop).\n
- Boundary markers: The skill lacks explicit instructions or delimiters that would direct the agent to ignore or isolate instructions found within the researched content.\n
- Capability inventory: The agent has the capability to write local files (via the docx skill), navigate and upload documents to DocuSign via the browser, and draft emails via the Gmail MCP.\n
- Sanitization: There are no specified sanitization or validation steps for content retrieved from external job postings before it is processed by the model.\n- Handling of Sensitive Information: The skill processes personally identifiable information (PII), such as candidate names, email addresses, and compensation details. While this is consistent with the skill's primary purpose, users should be aware that this data is handled in browser sessions and email drafts. The skill incorporates safety measures by requiring explicit user review and confirmation before any DocuSign envelope or Gmail draft is finalized.
Audit Metadata