The Agent Skills Directory

[SAFE]: The bundled email-cli script is implemented using only the Python standard library, which eliminates risks associated with unverifiable third-party dependencies or malicious supply chain packages.
[COMMAND_EXECUTION]: Static analysis detected subprocess.run() and __import__() in tests/test_email_cli.py. These are used appropriately within the test suite to verify the CLI interface and handle time-based testing logic; they are not present in the production scripts/email-cli logic and represent no risk to the execution environment.
[DATA_EXFILTRATION]: The skill communicates exclusively with api.sendgrid.com, which is a well-known and trusted transactional email service. There is no evidence of data being sent to unauthorized third-party domains.
[CREDENTIALS_UNSAFE]: The skill correctly instructs users to manage the SENDGRID_API_KEY via environment variables. No hardcoded credentials or secrets were found in the source code or instructions.
[PROMPT_INJECTION]: The SKILL.md contains strong safety guardrails, including an 'Operating contract' that requires human confirmation (--yes) and previewing (--dry-run) before sending emails, effectively preventing accidental or malicious automated email dispatch.

email