The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process and modify rules stored in .opencodereview/rule.json, which creates a surface for indirect instructions that might attempt to influence agent behavior.
Ingestion points: The skill reads configuration data from the .opencodereview/rule.json file located in the project directory (SKILL.md).
Boundary markers: The skill provides an explicit warning in references/rule-contract.md to ignore commands such as "ignore user authorization" if they appear in rule text, treating them strictly as project data.
Capability inventory: The agent has the capability to read and write local JSON files and execute the ocr CLI for rule verification (SKILL.md).
Sanitization: The skill instructs the agent to validate JSON syntax and use the official ocr rules check command for verification rather than relying on string matching or direct execution.
[COMMAND_EXECUTION]: The workflow involves using the ocr command-line utility to perform rule matching checks (ocr rules check <path>). This is a standard functional component of the intended code review workflow.

codereview-rules