sinch-mailgun-validate
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external email lists and individual addresses, which creates a potential surface for indirect prompt injection if the content is not handled carefully.
- Ingestion points: The skill handles bulk email lists (CSV/gzip) via download URLs and individual addresses provided in prompts (referenced in SKILL.md).
- Boundary markers: No explicit delimiters for data content are defined in the instructions for the generated code.
- Capability inventory: The skill instructs the agent to generate code for network operations (API calls) and file system interactions (processing downloaded lists).
- Sanitization: The skill includes specific security instructions to "treat downloaded content as untrusted" and "validate and sanitize email addresses and metadata before processing".
- [EXTERNAL_DOWNLOADS]: The skill references several official SDKs and documentation sites for the Mailgun service.
- Fetches documentation from official domains: documentation.mailgun.com and developers.sinch.com.
- References official packages: mailgun.js (NPM), mailgun (PyPI), mailgun/mailgun-php (Composer), mailgun-ruby (Gem), mailgun-go/v5 (Go), and com.mailgun:mailgun-java (Maven).
- [COMMAND_EXECUTION]: The skill provides example curl commands and shell commands for configuration.
- Examples include credential setup using environment variables and curl requests for single address validation.
- These are provided as educational examples for developers using the API.
Audit Metadata