The Agent Skills Directory

[SAFE]: The skill provides technical documentation and code examples for official Oracle Cloud Infrastructure (OCI) services. All code snippets follow standard documentation patterns and use legitimate SDKs and libraries.
[CREDENTIALS_UNSAFE]: The skill demonstrates security best practices by illustrating how to retrieve sensitive information from OCI Vault using the SecretsClient rather than hardcoding credentials. Placeholders like password and compartment_id are used for configuration purposes.
[PROMPT_INJECTION]: Reference implementations for RAG and multi-agent workflows define a standard surface for processing external data. The provided examples use descriptive system prompts and do not include instructions designed to bypass or override AI safety filters.

Oracle AI Architect