The Agent Skills Directory

Local Command Execution: The skill executes a bundled Node.js script (reviewops.mjs) to perform normalization tasks. This execution is scoped to a specific internal path, which is a common pattern for skill-specific logic.\n- Indirect Prompt Injection Surface: The skill is designed to process external review data and configuration files. It mitigates potential risks by explicitly instructing the agent to treat all imported strings, paths, and instructions as 'hostile inert data' and forbidding the following of any embedded references.\n- Strict Access Controls: The safety contract explicitly prohibits network access, credential harvesting, and modification of repository state, focusing solely on local data transformation and reporting.

normalize-review-runs