← ClaudeAtlas

codebase-auditorlisted

Scan a repository against curated coding standards and produce a structured audit report, issue set, refactor plan, and sprint-based remediation roadmap. Use when the user invokes /vibe.audit or asks to review the codebase against established rules. Operates in full autopilot mode from repository scan to sprint plan output.
Gladisintelligible706/vibe-driven-dev · ★ 0 · AI & Automation · score 68
Install: claude install-skill Gladisintelligible706/vibe-driven-dev
# Codebase Auditor ## Purpose This skill drives the `/vibe.audit` command. It turns curated coding standards into an operational review system. The goal is to let the AI coding agent perform a full repository audit in autopilot mode — scanning the codebase, selecting the right rule packs, comparing code patterns against those rules, generating structured issues with evidence, and producing an actionable sprint-based remediation plan. The output should feel like a **technical staff engineer audit**, not a lint dump. --- ## Core Rule Do not produce noise. Produce judgment. Every issue must be backed by evidence. Every finding must explain why it matters. Every remediation suggestion must be concrete and ordered. --- ## Autopilot Execution Flow Run these phases in order. Do not skip phases. Do not require user input between phases unless a halt condition is reached. --- ### Phase 1 — Understand the Repository Before applying any rule, understand what kind of project this is. **Inspect the following:** - root directory structure - `package.json`, `pyproject.toml`, `cargo.toml`, or equivalent manifest - key import patterns across source files - presence of frontend frameworks (React, Next.js, Vue, Svelte) - presence of backend frameworks (Express, FastAPI, Django, NestJS) - presence of AI or LLM integrations - presence of authentication systems - presence of background jobs, queues, or workers - presence of analytics or tracking hooks - presence of a test director