The Agent Skills Directory

[SAFE]: The skill provides documentation and instruction sets for performing architectural reviews. It does not contain executable code, initiate remote downloads, or include malicious prompt injections. It actively promotes security best practices by defining review axes that check for hardcoded credentials and unsafe coding patterns.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external data in the form of code diffs and project documentation, which constitutes an indirect injection surface.
Ingestion points: The skill explicitly identifies git diff output and CLAUDE.md content as its primary inputs in the 'Inputs' section.
Boundary markers: The instructions do not specify explicit delimiters or boundary markers for the ingested content, relying instead on the agent's contextual understanding.
Capability inventory: The skill possesses no active capabilities such as network access, file system modification, or process execution; its functionality is limited to generating a text-based review table.
Sanitization: No explicit sanitization or filtering logic is defined within the skill for the input data.

review