config-audit

You can audit Claude Code configuration by running a memory-backed two-phase workflow that applies a rubric to score over-engineering, redundancy, and token waste across hooks, skills, CLAUDE.md, and agents. It generates evidence-backed findings to guide remediation.

Over-engineering in Claude Code hooks and skills refers to redundant patterns, conflicting logic, and wasted tokens within configuration artifacts. A rubric-driven audit identifies these issues by quoting evidence from the configs and providing actionable optimization recommendations.

To check CLAUDE.md and agents for redundant configuration patterns, run an automated audit that uses rubric-driven scoring to surface conflicting patterns and wasted tokens. The analysis phase evaluates the artifacts and outputs remediation guidance to improve maintainability.

Yes, you can audit multiple Claude Code artifacts at once. The audit scans configured hooks, skills, CLAUDE.md, and agents simultaneously, applying a scoring rubric across all artifacts to identify redundancies and generate a unified findings report for remediation.

The best way to reduce token waste in Claude Code configurations is to use an automated audit that applies rubric-driven scoring to identify wasted tokens and redundant logic. It surfaces actionable optimizations with evidence quoting to improve cost efficiency.

No, the Claude Code config audit does not require external dependencies or components. It operates as a standalone skill using a memory-backed two-phase workflow that processes your existing configuration artifacts to generate scoring and remediation guidance.