cc-assess

cc-assess

Score and grade your Claude Code configuration against best practices.

cc-assess evaluates your Claude Code setup across 12 categories. Get your score, a visual radar chart, category deep-dives, and a shareable score card.

Quick Start

Option 1: Claude Code Plugin (Recommended)

# Install the plugin
claude plugin install cc-assess@Deway-AI/cc-assess

# Run the assessment
/assess

The plugin scans your config, scores it, copies the report to your clipboard, and sends you to claude-code-assess.web.app to see your full visual results.

Option 2: Export Script

No plugin? Run one command and paste the output at claude-code-assess.web.app:

bash <(curl -s https://claude-code-assess.web.app/export.sh) | pbcopy

Option 3: Manual Upload

Visit claude-code-assess.web.app and upload your config files directly (CLAUDE.md, settings.json, etc.)

What It Scores

Category	Weight	What's Checked
CLAUDE.md	25%	Exists, structured, uses imports, has standards and context
Settings & Permissions	20%	Permission rules, default mode, effort level
MCP Servers	15%	Configured, variety, team-shared
Security	12%	Deny rules, sandbox, no bypass mode
Hooks	8%	Automation, guardrails, lifecycle
Skills	5%	Custom skills with proper frontmatter
Agents/Subagents	5%	Custom agents with tool restrictions
Memory System	4%	Active memory with content
Rules Directory	2%	Path-specific rules
Plugins	2%	Installed from marketplace
IDE Integration	1%	VS Code or JetBrains
Advanced Features	1%	Teams, thinking, worktrees

Scoring

• Expert (90-100): Fully optimized power user
• Advanced (75-89): Strong setup, room for improvement
• Intermediate (50-74): Good foundation, missing key features
• Beginner (25-49): Basic setup, many features unused
• Getting Started (0-24): Minimal configuration

How It Works

The plugin reads your local config files (CLAUDE.md, settings.json, .mcp.json, skills, agents, hooks, rules, memory, plugins) and scores each against documented best practices. Everything is read-only — nothing is modified.

The web app runs the same scoring logic entirely client-side. No data is sent to any server.

Differentiator

Anthropic's official claude-code-setup plugin is a recommender — it analyzes your codebase and suggests what Claude Code features to add. cc-assess is an assessor — it scores how well your existing configuration is set up. They're complementary tools.

License

MIT

Credits

Built by Deway