The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The documentation mentions installing official libraries like mailgun.js and form-data from the public registry.
[COMMAND_EXECUTION]: Provides standard examples for using curl and npm for API interaction and environment setup.
[PROMPT_INJECTION]: The skill is designed to handle external data from emails and webhooks, which presents a surface for indirect prompt injection. It proactively addresses this with defensive instructions.
Ingestion points: External email content and webhook payload data.
Boundary markers: Explicit instruction to avoid following URLs from untrusted domains.
Capability inventory: Ability to send emails and query logs via API, and access to local environment configurations.
Sanitization: Instructions restrict automated URL fetching to trusted first-party domains.
[SAFE]: Guidelines to retrieve API keys from environment variables or .env files follow standard security protocols for secret management in developer tools.

sinch-mailgun