sinch-mailgun-optimize
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-supplied data, including domain names, IP addresses, and email body content, which are used to construct and execute API requests.
- Ingestion points: User inputs for domain registration (
POST /v1/inboxready/domains), IP registration (POST /v1/inboxready/ip_addresses), and inbox placement tests (POST /v4/inbox/tests). - Boundary markers: No explicit delimiters or instructions are provided to the agent to treat user-supplied strings as data only or to ignore potential commands embedded within that data.
- Capability inventory: The skill uses
curlto perform network operations against the Mailgun/Sinch API endpoints. - Sanitization: There are no instructions for validating, escaping, or filtering the user-provided inputs before they are interpolated into the shell commands or API payloads.
Audit Metadata