principal-hierarchy-audit

Audit system prompts against the Anthropic constitutional hierarchy by providing a system prompt or plugin file path to generate a structured report. The report identifies compliance violations, classifies severities as RED, YELLOW, or GREEN, and returns recommended fixes.

A principal hierarchy audit evaluates system prompts and plugin configurations to identify instructions that conflict with Claude's training, attempt to weaponize the AI, violate user protections, or exceed operator permission boundaries. It produces a structured machine-readable compliance report.

Check plugin configurations for safety violations by running an audit that detects identity deception, hard constraint breaches, and operator boundary exceedances. The audit references specific constitutional principles and provides actionable compliant alternatives for each violation.

No external API calls are required to run a prompt compliance audit. The tool operates locally by analyzing the provided system prompt or plugin file path and returning a self-contained machine-readable report with constitutional references and remediation guidance.

Classify prompt safety violations by severity using an audit that generates RED, YELLOW, and GREEN classifications. The audit maps violations to implicated constitutional hierarchy levels and outputs a structured report with clear remediation guidance for operators.

Yes, you can audit SKILL.md frontends for constitutional compliance. The audit evaluates SKILL.md frontends, plugin prompts, and system prompts, producing a structured report with constitutional references and compliant alternatives for any identified violations.