pre-send-qa
Pass
Audited by Gen Agent Trust Hub on May 2, 2026
Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted email HTML and text content from external Email Service Providers or direct user input, creating a surface for indirect prompt injection.\n
- Ingestion points: Campaign content fetched via ESP-specific MCP tools and user-pasted HTML strings in Mode B.\n
- Boundary markers: The instructions lack delimiters or explicit directives for the agent to ignore instructions embedded within the email content being analyzed.\n
- Capability inventory: The skill environment allows tools such as
Bash,Write, andWebFetch, which could be exploited if an injection successfully influences agent behavior.\n - Sanitization: No sanitization or filtering of the ingested content is specified before the analysis steps.\n- [DATA_EXFILTRATION]: The skill utilizes the
WebFetchtool to perform network requests (HEAD checks) on arbitrary URLs extracted from the email body. While the intended purpose is link validation, this involves automated network operations to non-whitelisted external domains based on untrusted input data.
Audit Metadata