The Agent Skills Directory

[SAFE]: The skill serves as a reference and design specialist for AI/ML systems. It provides documentation, architectural patterns, and evaluation frameworks rather than executable code.
[SAFE]: No prompt injection patterns were detected. The skill contains defensive guidelines against prompt injection (referencing OWASP LLM Top 10) and advocates for instruction/data separation.
[SAFE]: No malicious data exfiltration or credential harvesting patterns were found. Mentions of PII handling and security reviews are centered on defensive design and compliance.
[SAFE]: No obfuscation techniques, such as hidden Base64 strings, zero-width characters, or homoglyph attacks, were identified in the text or metadata.
[SAFE]: The skill does not perform external downloads or remote code execution. It references well-known platforms and tools (Anthropic, OpenAI, Pinecone, etc.) for design and integration purposes.
[SAFE]: No privilege escalation or persistence mechanisms were found. The skill operates as a high-level design assistant without requesting or managing system permissions.

oracle