codebase-review

Audits an entire repository for DRY violations, dead code, and naming drift using parallel reviewer agents.

Updated May 20, 2026
One-click install
npx skills add https://github.com/TechCorp25/kingdom --skill codebase-review-techcorp25
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: codebase-review
Source: https://github.com/TechCorp25/kingdom/tree/main/.claude/skills/codebase-review
Command: npx skills add https://github.com/TechCorp25/kingdom --skill codebase-review-techcorp25

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve? Per-task code reviews only see one branch or diff at a time, so cross-module duplication, accumulated complexity, and naming drift go unnoticed. This Skill performs periodic whole-repo audits that catch the debt incremental reviews miss. ## Core Features & Use Cases - Two execution modes: Single mode runs one reviewer per top-level directory plus cross-scope reconciliation; team mode runs three independent full-repo reviewers with peer cross-verification and confidence tiers like [VERIFIED 3/3] and [DISPUTED]. - Five review categories: DRY, YAGNI, Simplicity & Efficiency, Refactoring Opportunities, and Consistency, each with Critical/High/Medium/Low severity ratings and file:line citations. - Automatic routing of findings: Inline fixes are handed to draft-plan and plan-review automatically, while complex fixes become GitHub issues or plans after user confirmation. - Use Case: After a long development sprint, run a team-mode audit to produce a dated report in docs/reviews/ ranking every finding by severity, then auto-generate fix plans for the simple ones. ## Quick Start Ask the assistant to run /codebase-review on the repository to audit the whole codebase and produce a severity-ranked findings report.

Frequently Asked Questions about codebase-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a whole-repo code quality audit?▼

Invoke /codebase-review with no arguments to audit the entire repository, or pass a path to scope it to one directory. The skill dispatches parallel reviewer agents per top-level directory, reconciles cross-directory findings, and writes a severity-ranked report to docs/reviews/.

What is the difference between single mode and team mode in codebase review?▼

Single mode runs one reviewer per directory plus a cross-scope reconciliation pass at roughly 1x token cost. Team mode runs three independent full-repo reviewers with peer cross-verification and confidence tiers at roughly 3x token cost, surfacing findings a single reviewer would miss.

When should I use codebase-review instead of a branch or diff review?▼

Use it for periodic audits, pre-refactoring assessments, or after long development phases when quality may have drifted. It is not for branch reviews (use implementation-review) or diff-only review (use /simplify), since those only see one change at a time.

Why does team mode downgrade to single mode?▼

Team mode requires the CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1 environment variable. If it is unset, the mode-detect helper emits a warning and downgrades to single mode automatically, so the run still proceeds without an error.

What happens to the findings after a codebase review completes?▼

Findings are deduplicated, ranked by severity, and grouped by overlapping file sets. Simple inline fixes are routed automatically to draft-plan and plan-review for execution, while complex fixes prompt you to create GitHub issues or write plans per group.