The Agent Skills Directory

[SAFE]: The skill consists entirely of instructional markdown files and configuration schemas. No security vulnerabilities or malicious patterns were detected.
[PROMPT_INJECTION]: The documentation describes tools that ingest external data (API responses and DTMF keypad entries), which constitutes an indirect prompt injection surface. Ingestion points: External API responses (defined in references/tool-schemas.md) and DTMF input (references/dtmf.md). Boundary markers: Not explicitly defined in schemas, but instructions recommend clear agent confirmation steps and read-back procedures. Capability inventory: Call transfers and HTTP request execution (SKILL.md, references/transfer-call.md). Sanitization: Not present in the architectural design.
[CREDENTIALS_UNSAFE]: The documentation provides examples for configuring API authentication. All keys shown (e.g., 'sk_live_abc123', 'helpdesk_key_789') are placeholders used for illustrative purposes.

setup-tools