Code Analyzer Suite

This repository contains two related skills:

• code-analyzer-suite: the original version, designed for structured multi-dimensional review with manual multi-window fallback.
• code-analyzer-auto: the auto-orchestration version, designed for runtimes that can dispatch sub-agents or worker threads automatically.

Use the original skill when you want explicit worker prompts or need to support older environments. Use the auto version when the runtime can orchestrate parallel workers directly.

Which Version Should You Use?

Variant	Use it when	Avoid it when
code-analyzer-suite	You want explicit review dimensions, visible worker prompts, or a manual fallback path	Your runtime can already dispatch review workers automatically and you want the orchestration version
code-analyzer-auto	Your runtime can orchestrate sub-agents or worker threads and you want consolidated auto-dispatched review	You need the review flow to stay manual, copy-pasteable, or compatible with older environments

The subject matter is the same in both versions. The difference is the execution model: manual or mixed review orchestration versus auto-dispatched review orchestration.

Workflow Generations

This repository participates in a broader workflow lineage:

1. Generation 1: the manual workflow stack code-analyzer-suite plus parallel-decomposer-skill, grill-me, and iterative-implementation-review
2. Generation 2: the auto-capable workflow stack code-analyzer-auto plus parallel-decomposer-auto, grill-me, and iterative-implementation-review-auto
3. Final integrated distribution implementation-workflows, which packages both the standard and auto paths into one workflow suite

In that lineage, this repository is the review family for both generation 1 and generation 2.

Skills In This Repo

code-analyzer-suite

• Focused multi-dimensional review across Security, Performance, Code Quality, Architecture, and Logic Verification
• Structured findings-first output
• Manual or semi-manual parallel execution
• Good fit for environments where worker prompts must be visible and copy-pasteable

code-analyzer-auto

• Automatic dispatch to sub-agents or worker threads when the runtime supports it
• Shared orchestration brief for all workers
• Consolidation-oriented output for orchestrator agents
• Manual fallback prompts only when automatic execution is unavailable
• Trigger boundary that keeps ordinary single-agent reviews on the original skill

Features

• 5 analysis dimensions: Security, Performance, Code Quality, Architecture, Logic Verification
• Parallel review support for both automatic worker orchestration and manual multi-window execution
• Structured output with consistent severity ratings
• Language support for Python, JavaScript/TypeScript, Java, Go, Rust, SQL, and more
• Consolidated reports that merge dimension results into one review

Related Skill

For general-purpose task decomposition, use parallel-decomposer-skill.

For the workflow layer that uses this repository as a review gate, see iterative-implementation-review.

If you want the full packaged workflow, see implementation-workflows.

Relationship To The Other Repositories

This repository answers the question:

• how should changed code be reviewed across multiple dimensions?

The companion repositories answer different questions:

• parallel-decomposer-skill: how should work be split safely?
• iterative-implementation-review: how should implementation, critique, repair, and verification be looped together?

That separation is intentional. This repository should stay focused on evidence-based review rather than owning the whole implementation workflow.

Repository Roles At A Glance

Repository	Primary job	Typical output
parallel-decomposer-skill	split work safely	task cards or worker specs
code-analyzer-suite	review changed code	findings and severity-ranked risks
iterative-implementation-review	keep looping until the implementation survives review	repaired implementation plus verification status

Installation

Quick Install

./install.sh

The quick installer targets the original code-analyzer-suite. Install code-analyzer-auto as a separate skill directory when your runtime supports auto-discovery of multiple skills.

Install Auto Skill

cp -R code-analyzer-auto ~/.codex/skills/code-analyzer-auto
cp -R code-analyzer-auto ~/.agents/skills/code-analyzer-auto

Use the first path for Codex and the second path for universal agent runtimes. On Windows, copy code-analyzer-auto to the matching skills directory for your tool.

Manual Install by Platform

Claude Code

cp -R code-analyzer-suite ~/.claude/skills/code-analyzer-suite

GitHub Copilot (project-level)

cp -R code-analyzer-suite .github/skills/code-analyzer-suite

Cursor (project-level)

cp -R code-analyzer-suite .cursor/skills/code-analyzer-suite

Windsurf

cp -R code-analyzer-suite ~/.codeium/windsurf/skills/code-analyzer-suite

Universal

cp -R code-analyzer-suite ~/.agents/skills/code-analyzer-suite

Usage

Original Skill

/code-analyzer Review this Python API for security and performance issues
/code-analyzer Analyze the authentication module for vulnerabilities and logic errors
/code-analyzer Check this React component for performance, accessibility, and best practices
/code-analyzer Audit the entire codebase for security risks
/code-analyzer Review src/auth/login.ts for all dimensions

Auto Skill

Use $code-analyzer-auto to review this pull request by dispatching Security, Performance, and Logic workers and consolidating the findings.

Repository Structure

code-analyzer-suite/
|-- SKILL.md
|-- AGENTS.md
|-- code-analyzer-auto/
|   |-- SKILL.md
|   |-- AGENTS.md
|   |-- agents/openai.yaml
|   |-- assets/orchestration-brief-template.md
|   |-- references/orchestration-guide.md
|   `-- evals/code-analyzer-auto.eval.md
|-- agents/
|-- assets/
|-- references/
|-- scripts/
`-- evals/

License

MIT