agent-reviewer

Validate Claude Code agent definitions by checking frontmatter, description, system prompt, tool scope, and model tier. This automated review enforces spec compliance and quality standards through detailed reporting, catching issues like incorrect tool scope or vague descriptions before deployment.

Agent compliance auditing verifies that AI agent definitions adhere to strict correctness, quality, and consistency standards. It performs automated checks across a 5-dimension framework to identify areas for improvement, ensuring only high-quality, reliable agents are deployed into your codebase.

Audit existing agents by validating their definition files against a comprehensive framework covering frontmatter correctness, description quality, system prompt structure, tool scope appropriateness, and model tier justification. This process identifies spec violations and areas for improvement through detailed reporting.

Yes, agent validation optionally uses ecosystem.json for cross-referencing. The process requires access to agent definition files and leverages this ecosystem data to enforce spec compliance and verify quality standards across your AI development environment.

Automated agent definition validation is limited to checking structural and spec compliance across five fixed dimensions: frontmatter, description, system prompt, tool scope, and model tier. It cannot assess the logical correctness or runtime behavior of the agent's underlying prompt engineering logic.

Yes, agent review checks tool scope appropriateness and model tier justification as part of its comprehensive 5-dimension validation framework. It enforces spec compliance by catching potential issues like incorrect tool permissions or unjustified model assignments in agent definitions.