The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it reads external SMS content.
Ingestion points: Reading SMS messages via read_phone_inbox in SKILL.md.
Boundary markers: None specified.
Capability inventory: No high-risk capabilities like file system or shell access are requested in this skill.
Sanitization: No sanitization of incoming SMS content is mentioned.
[SAFE]: The skill follows security best practices for credential management by instructing users to use dedicated tool calls (get_credential, store_credential) instead of hardcoding secrets or placing them in prompts.

phone-verification