The Agent Skills Directory

[SAFE]: The skill acts as a configuration and instruction set for an evaluation workflow. It utilizes vendor-owned tools (Ouroboros) associated with the author (Q00) to perform standard development tasks like linting and testing.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface inherent to its function of analyzing external content.
Ingestion points: The skill ingests untrusted content through the artifact and session_id parameters passed to the ouroboros_evaluate tool in SKILL.md.
Boundary markers: There are no explicit delimiters or specific 'ignore instructions' warnings defined in the prompt instructions to separate the artifact content from the agent's logic.
Capability inventory: The associated ouroboros_evaluate tool is capable of executing build commands and running tests, which involves subprocess execution.
Sanitization: The skill instructions do not specify any sanitization or validation of the artifact content, relying on the internal safety mechanisms of the MCP tool.

evaluate