Compare
Claude Code insights, ccusage, WakaTime and the alternatives: what each one tells you
The short answer
Each tool answers a different question. ccusage, /usage and Claude Code Usage Monitor count tokens, cost and limits. /insights and AI Engineer Coach look at how you use the AI and suggest better habits and setup. WakaTime counts your coding time. DX and review bots report on a team's output and code. who-said-yes counts which agent actions you approved yourself. Sense is the one built to show whether your own judgment is getting better.
- Spend and limits: ccusage, /usage, Claude Code Usage Monitor.
- How you use the AI: /insights, AI Engineer Coach.
- Time and team output: WakaTime, DX and engineering analytics, review bots.
- Your own calls and judgment: who-said-yes and Sense (both ours).
Every fact checked at its source on 9 November 2026
Start from your question
- What did this cost me? Run
npx ccusage@latestor/usagein Claude Code. - Will I hit my limit before I finish? Claude Code Usage Monitor.
- How do I use Claude Code, and where does it go wrong?
/insights. - What habits do I have across Copilot, Claude Code and Codex? AI Engineer Coach.
- How much time do I spend coding, and how much is AI? WakaTime.
- How is my team doing with AI? DX and other engineering analytics, which your company buys.
- Is this pull request safe to merge? A code review bot, then you.
- How many of my agent's actions did I approve myself? who-said-yes.
- Am I still getting better at the work as agents do more of it? Sense.
Every tool in one table
What each tool shows, who it is for, whether it makes the AI better or you better, where your data goes, and what it costs.
| Tool | What it shows | Who it is for | AI better or you better? | Where your data goes | Price |
|---|---|---|---|---|---|
| /insights (Claude Code) | A report on what you work on, friction points such as misunderstood requests or buggy code, and suggestions | You, in Claude Code | The AI: suggestions for using Claude Code more effectively | Analysis runs through your own Claude account; report saved on your computer | Included; uses your plan's tokens |
| /usage (Claude Code) | Session cost, plan limits, what uses your plan, and changes to cut usage | You, in Claude Code | Your setup and spend | Computed from local history on this computer | Included |
| ccusage | Tokens and estimated cost by day, week, month and session | You, across Claude Code, Codex and 16 other agents | Neither: it counts spend | Reads local logs in your terminal | Free, open source (MIT) |
| Claude Code Usage Monitor | Live burn rate and a forecast of when you hit your limit | You, in Claude Code | Neither: it counts limits | Local, in your terminal | Free, open source (MIT) |
| AI Engineer Coach | Practice scores, 45 anti-pattern rules on prompts, sessions and context, code volume, quizzes from your usage | You, in VS Code or the GitHub Copilot app; reads Copilot, Claude Code, Codex, OpenCode and Xcode logs | Mostly the AI: how you prompt and set it up; its quizzes test you | Local; some optional features call GitHub Copilot's model | Free, open source (MIT); you build it yourself |
| WakaTime | Coding time by project, language and editor; AI use, spend and AI vs human code on paid plans | You, or a team | Neither: it reports time and AI use | Plugins send file paths, project names and editor activity to WakaTime; never file contents | Free (1 week of history); $9 to $14 a month; teams from $21 per developer |
| DX and engineering analytics | Delivery, developer surveys, AI-generated code by commit, PR, team and repo | Engineering leaders | Neither: it reports a team's output and AI impact | Your company's tools and surveys, in the vendor's platform | Not public; sold by demo |
| Code review bots (CodeRabbit and others) | Review comments on pull requests | Teams that ship code | The code: problems caught before merge | Your code goes to the vendor for review | CodeRabbit: $24 to $72 per developer a month billed yearly; free on public repos |
| who-said-yes (ours) | How many agent actions you approved yourself, your longest run with no message, edits you never saw | You, in Claude Code and Codex | Neither: a count about you, with no advice | Local; no network code, no account | Free, open source (MIT) |
| Sense (ours) | Whether your own judgment with AI is getting better, with the moments behind it in your words | You, in Claude Code, Codex, the Claude and ChatGPT apps and Antigravity | You: practices chosen from your own chats, with credited lessons | Your record stays on your computer; usage counts reach KORTURE.ai | Free research preview |
From each tool's own docs, README or pricing page, read 9 November 2026. Prices in US dollars. KORTURE.ai makes who-said-yes and Sense.
The AI better, or you better?
Most of these tools are about the AI or the bill. /insights suggests features to try and fixes to your setup (Claude Code docs). /usage recommends changes that cut what you use (Claude Code docs). Review bots look at the code, and DX reports to leaders on a team's output; Atlassian bought DX in 2025 (DX).
AI Engineer Coach comes closest to looking at you. It is an open source project by Microsoft employees, not an official Microsoft product. It scores your practice, flags 45 habits across prompt quality, session hygiene, code review, tool mastery and context, and builds quizzes from your own usage (AI Engineer Coach). Its rules are mostly about how you drive the AI.
How fast it feels is a poor guide either way. In METR's 2025 study, experienced open source developers expected AI to make them 24% faster and afterwards believed it had made them 20% faster. Measured, they took 19% longer (METR).
Looking for a ccusage alternative or a WakaTime alternative because you want to know about yourself, not your tokens or hours? That is the gap Sense is built for. How KORTURE.ai measures it: Sense reads your own chats on your computer and shows the moments behind each reading in your own words, such as when you made the call before asking and when you took the first answer (what Sense is).
Beyond counts
See whether your own judgment is getting better
Sense reads your own chats with AI on your computer and shows how you are really doing, with the moments behind it in your own words. Tell us your computer and where you work with AI. If Sense runs there today, your copy arrives by email.
Get SenseResearch preview: free, on Apple silicon Macs, and on Windows as a test.
Sources
- Manage costs effectively: /usage and /insights
- Commands
- ccusageSources it reads, reports, MIT licence
- Claude Code Usage Monitor
- AI Engineer CoachFeatures, logs it reads, privacy, disclaimer
- WakaTime pricing
- WakaTime FAQWhat plugins send
- DX
- DX is joining Atlassian
- CodeRabbit pricing
- who-said-yes
- What is KORTURE.ai Sense?
- Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity