How AI-native are you, your team, your next hire?

AIQ Rank scores how good you actually are at Claude Code, Codex CLI, OpenCode, Cursor, and Cowork. Real activity. Not token counts. Your data stays on your machine.

60-sec scan · Free for individuals · Free private leaderboards

Get my Free AIQ For teams See today's leaderboard →

AIQ · Percentile

/ aiqrank

97 %

Rank S · Peak

Skill breakdown

Customization 94

Parallel Agents 88

Tool Breadth 81

Planning 95

Multi-Tasking 78

Custom Skills 85

Sample profile 30-day window

Why do this?

Are you using AI to the max, or just counting tokens?

We built AIQ Rank because we kept asking ourselves the same question: are we actually using our AI tools well, or just using them? We'd watch other people casually drop fifteen agents in parallel or wire up MCP servers we'd never heard of, and wonder what we were missing.

An AIQ score answers it. You see how your tool diversity, agent orchestration, skills, and context leverage compare to everyone else who's been scanned. No guessing, no vibes. If you're crushing it, you'll know. If there's room to grow, you'll see exactly where.

Think of it as a fitness tracker for your AI craft. The rank is the brag; the breakdown is the practice.

What we measure

Eleven dimensions of AI coding skill.

Real proficiency with Claude Code, Codex CLI, and OpenCode shows up across eleven dimensions. Eight apply to every tool. Three are tool-specific bonuses. Each is read from your actual session activity, not token counts.

Customization

Have you shaped the tool to your workflow?

CLAUDE.md, AGENTS.md, custom slash commands, hooks.

Parallel Agents

Can you orchestrate multiple agents at once?

Subagents, parallel runs, specialized agent definitions.

Background Work

Do you delegate work to run while you're elsewhere?

Sessions that leverage context to act on your behalf.

Tool Breadth

How wide is your AI workflow?

MCP servers, skills, plugins wired into your environment.

Planning

Do you plan before acting?

Plan mode usage, structured workflows, /spec and /plan.

Repetition

How deeply do you engage with your skills?

Total skill invocations across the scoring window.

Custom Skills

Have you built reusable workflows of your own?

Skills, commands, and agents you've written yourself.

Multi-Tasking

Are you treating AI like a team, not a chat window?

Multiple Claude Code, Codex CLI, OpenCode sessions in parallel.

Thinking Mode

Codex

Are you using deep reasoning when it counts?

Codex CLI reasoning engagement on hard tasks.

Schedule Reliability

Cowork

Are your scheduled agents firing reliably?

Cowork sessions running on their planned cadence.

Queue Discipline

Cowork

Do you follow through on the work you queued?

Cowork queue completion across the window.

Read the full methodology in How to measure Claude Code proficiency (beyond token counts) .

Set up in seconds.

One plugin for Claude Code CLI, Cowork, and Codex. No config, no account required to scan.

Private by default.

Transcripts never leave your machine. You review exactly what uploads before it does.

Built to share.

Score cards for the timeline. Profile URL. Leaderboard spot. Bragging rights, backed by data.

Setup · one prompt

Paste one prompt. Done.

Works in Claude Code, Cowork, Codex, or Cursor — the AI identifies its own runtime and runs the right install for you.

Copy and paste into your AI assistant

Prefer to install it yourself? See manual install for Claude Code, Cowork, Codex, or Cursor →

For companies

Is your team using AI well, or just counting tokens? Same for your next hire.

Spin up a private leaderboard in 60 seconds. See who's actually good at Claude Code, Codex CLI, and OpenCode across engineering, product, and marketing. Free during launch.

Employees

See how your engineers stack up internally, by tool, skill, and team. Spot who's pulling AI to the max and who needs a hand.
Candidates

Send the setup link plus a handle to use. They run the scan, you see their AI fluency on your private leaderboard before the interview.

Admins see tool, skill, and MCP names for company members. Never prompts, code, or conversation contents. Members opt in with a join token.

◆ Create a leaderboard

60 seconds, free during launch.

1. Sign in with your email (or Google / GitHub).
2. Pick a name and a slug.
3. Copy the share command to your team.

Create a private leaderboard →

Already signed in? Skip to the form →

FAQ

Questions,
answered.

Didn't find yours? Ask us.

Is it free?

For individuals, yes. No credit card, no paywall, no "pro plan." For companies, AIQ Rank is in private beta: free to join right now, paid plans land later.

Is my data secure?

Your transcripts never leave your machine. The plugin scans locally and uploads aggregated counts plus the names of tools, skills, MCP servers, and agents you use. Those names are visible only to you on your own private dashboard. They are never shown on the public profile, leaderboard, or share card. (One exception: if you associate your account with a company, that company can see those names for its associated accounts.) We never see your prompts, code, conversations, or file contents. Full details on the privacy page.

How often does my score update?

Every time you run /aiqrank the score recomputes from your last 30 days of activity across every supported tool. Re-run weekly to watch your rank move, or any time you want a fresh snapshot.

How do you make money?

Companies pay for the private corporate version that lets them see how their engineers, teams, and candidates stack up. Free during the current private beta. We may also enable opt-in job-recruiting matches for top-ranked individual users. Your individual data is never sold.

Does this work with tools other than Claude Code?

Yes. The plugin reads Claude Code, Claude Cowork (autonomous local-agent-mode sessions), OpenAI Codex CLI rollouts, OpenCode, and Cursor. Your score combines them into one All-tools rank, with per-tool drill-ins. Copilot, Windsurf, and others may follow; let us know which matters most.

Does it track Claude Code on the desktop or web?

Partially. The plugin captures every Claude Code CLI session and every Local Agent Mode (Cowork) session on your machine. That covers most of what people call "Claude Code on the desktop." Cloud-hosted sessions (the kind that run in Anthropic's cloud and stream to claude.ai, the desktop chat tab, or the iOS app) live entirely server-side and aren't exportable yet, so they don't count toward your score today. We're watching that gap and will fold them in if Anthropic ships an export.

Can I share my score?

That's kind of the point. You get a profile URL, a shareable score card sized for X / LinkedIn / Bluesky, and an optional leaderboard spot. Bragging rights, backed by data.

What does the score actually measure?

Eleven dimensions in total. Eight core that apply to every tool: customization, agent orchestration, background work, tool breadth (skills + MCPs), planning, repetition, custom skills, multi-tasking. Plus three tool-specific bonuses: reasoning engagement (Codex), schedule reliability (Cowork), queue discipline (Cowork). Each dimension gets a percentile; the AIQ is the weighted composite. Full methodology lives in the docs.

Ready to see your AIQ?

Paste the one prompt above. Takes 60 seconds. Nothing uploads without your say-so.