The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines procedures for interacting with the oracle CLI tool and MCP consult tool. It utilizes commands such as oracle --engine, oracle --render, and oracle --help --verbose to perform model inquiries and preview data bundles.
[INDIRECT_PROMPT_INJECTION]: The skill processes project files (e.g., package.json) and provides them to external AI advisors. Ingestion points include local files passed via the --file argument. Boundary markers are not specifically detailed, but the skill emphasizes that the calling agent must manually check advice. The skill includes a clear instruction to exclude credentials from any task context sent to the advisor.
[DATA_EXFILTRATION]: The skill transmits repository content to external AI providers through API and browser-based engines. This is a primary function of the tool, and the skill includes specific constraints to prevent unauthorized data exposure, such as excluding credentials and verifying transport availability before fallback.

oracle-advisor