The proficiency platform for Claude Code

You're using Claude Code.
But are you using it well?

Most developers never find out. No score, no feedback, no way to know if you're in the top 10% or wasting half your tokens. We changed that.

Free tier forever · No credit card · 2 min setup

What we found

Early data from Claude Code sessions
reveals a clear pattern.

3x+

more tokens burned by developers who skip CLAUDE.md

Most

developers never use plan mode — the single biggest efficiency lever

Top 10%

get dramatically more value from the same Claude subscription

The gap isn't talent. It's habits. And habits can be coached.

Introducing CPR

Your Claude Code Proficiency Rating

A single number from 0 to 2000 that shows how effectively you work with Claude Code. Like ELO for chess — but for AI-assisted development.

1,342Proficient
Practice Score
72
Outcome Score
65
Challenge Score
58
Foundation 0–799Practitioner 800–1199Proficient 1200–1599Expert 1600–1899Master 1900+

Public profile. Embeddable badge for your GitHub README. Put it on your resume.

The coaching stack

Four layers working together
to make you measurably better.

01 — Dashboard

Smart Dashboard

Your CPR score front and center, with a 7-day trend sparkline, category breakdown, and three personalized coaching cards that rotate as you improve.

1,342CPR
Coaching

Add file references to prompts for 2x better context

Coaching

Try plan mode for multi-step tasks — your biggest score lever

Terminal

$ claude "refactor the auth module"

⠋ Working...

💡 Tip: Add file paths to your prompt for better context.

Try: claude "refactor src/lib/auth.ts"

02 — Terminal

Terminal Coach

Contextual nudges right in your workflow. When you miss a best practice, a subtle inline hint shows exactly what to do differently — no tab switching needed.

New: Claude Code now supports plan mode chaining

2 hours ago · claude.ai/blog

Community: Top CLAUDE.md patterns from 500+ repos

5 hours ago · github.com

Your score impact: +15 if you adopt compact mode

Personalized · Based on your usage

03 — Intelligence

Ecosystem Intelligence

An AI pipeline monitors the Claude Code ecosystem — new features, community patterns, best practices — and delivers a personalized feed of what matters to your score.

Refactor Challenge #421200 rated

Extract a reusable auth middleware from a monolithic Express handler using Claude Code.

⏱ 25 min · 3 filesStart →
04 — Challenges

Rated Challenges

Timed coding sessions rated like Codeforces. Solve real problems with Claude Code, earn Challenge Score points, and climb the leaderboard.

For engineering teams

Your team has the same gap. Multiplied.

You bought Claude Code licenses for 20 engineers. How many are using it well? How many are burning tokens on bad prompts? You have no idea. Until now.

  • See who's ramping up and who needs help
  • Track adoption velocity across the org
  • Measure ROI of your Claude Code investment
  • Compare team proficiency to industry benchmarks

Team Overview

1,180
Avg CPR
+12%
Growth
18/20
Active
1Sarah C.
1420
2Mike R.
1380
3Ana K.
1195
+ 17 more

How it works

Two minutes to your first score

01

Install the plugin

Run a single command to install the monitoring plugin. It hooks into Claude Code's telemetry — zero config, no code changes, no performance impact.

curl -sL prompt1.spectatr.ai/install.sh | sh
02

Use Claude Code normally

The plugin silently measures best practice signals as you work. Prompt quality, tool usage, workflow patterns — all tracked automatically.

03

Get coached to improve

Dashboard shows your score, coaching cards, and a personalized path to improvement.

Trust & Security

Built for developers who care about privacy

Coaching, not surveillance. You own your data.

No source code collected

The plugin collects practice signals only — never your code, prompts, or Claude's responses.

Encrypted end-to-end

TLS 1.3 in transit, AES-256 at rest. Your data is protected at every layer.

Zero performance impact

Hooks run async after operations complete. Your Claude Code experience stays fast.

Open source plugin

Audit every line. The plugin source is fully open — see exactly what runs on your machine.

Pricing

Start free. Grow from there.

Every plan includes the plugin, CPR score, and public profile.

Free

$0forever

Get started with your CPR score

  • CPR score + public profile
  • GitHub badge
  • 3 challenges / month
  • Basic dashboard
Get started

Individual

$19/month

Full coaching and intelligence

  • Everything in Free
  • Smart Dashboard + coaching
  • Terminal Coach nudges
  • Ecosystem intelligence
  • Unlimited challenges
Start free trial

Enterprise

Customcontact us

Full platform with SSO and interviews

  • Everything in Team
  • SSO / SAML
  • Custom challenge pools
  • 90-day history + SLA
  • Interview platform access
Contact sales

14-day free trial on all paid plans · No credit card required · Cancel anytime

FAQ

Frequently asked questions

What data does the plugin collect?+

The plugin tracks best practice signals — prompt quality metrics, tool usage patterns, workflow efficiency, and security practices. It does NOT log prompt content or your code. All data is aggregated into practice scores.

Will it slow down Claude Code?+

No. The plugin hooks run asynchronously after Claude Code operations complete. There is zero impact on response time or performance.

Is this a monitoring tool or a coaching tool?+

Coaching. Every developer sees their own proficiency score and personalized recommendations. The goal is to help developers improve, not to surveil them. Team dashboards show aggregate trends, not individual prompt logs.

What platforms are supported?+

The plugin works on macOS and Linux. Windows support via WSL is coming soon. It requires Claude Code CLI (the VS Code extension is not yet supported).

How is the proficiency score calculated?+

We measure best practices across prompt quality, workflow efficiency, and security. Each practice is weighted and combined into a 0-100 composite score. The score goes up as you build better habits.

What happens after the free trial?+

After 14 days, you'll need to subscribe to continue accessing insights. Your historical data is retained for 30 days after trial expiry so nothing is lost if you decide to subscribe.

What is CPR?+

CPR stands for Claude Code Proficiency Rating. It's a score from 0 to 2000 that measures how effectively you work with Claude Code, based on three pillars: Practice Score (your daily habits), Outcome Score (your results), and Challenge Score (timed coding challenges). Tiers range from Foundation (0–799) to Master (1900+).

Is the free tier really free?+

Yes. Your CPR score, public profile, GitHub badge, and 3 challenges per month are free forever. No credit card required. Paid plans add the coaching dashboard, terminal nudges, ecosystem intelligence, and unlimited challenges.

How do challenges work?+

Challenges are timed coding sessions where you solve real problems using Claude Code. They're rated like Codeforces — your Challenge Score goes up when you solve harder problems faster. 3 challenges per month are free; unlimited with a paid plan.

Find out how good you really are with Claude Code

Install the plugin. Get your CPR score. Start improving today.

Get your CPR score — freecurl -sL prompt1.spectatr.ai/install.sh | sh