We launched on Product Hunt today — here's what we shipped

Today I launched Claude Token Dashboard on Product Hunt.

👉 https://www.producthunt.com/posts/claude-token-dashboard

Here's what happened in the last two weeks.

---

Where it started

I had no visibility into what Claude Code was doing.

The cost was showing up in my billing. But which project?
Which session? What was the AI actually doing?

Turned out all the data was sitting on my machine the whole time —
~/.claude/projects/ — every session logged as JSONL.

So I built a dashboard to read it locally.

---

What shipped (v0.1.x → v0.2.2)

Token & Cost Visibility
Token usage and dollar cost by project.
Plan-aware: Pro/Max shows token volume, API shows dollar cost.

Today's Summary + Month Projection
Today's cost, tokens, actions, active projects — at a glance.
Plus: "based on this month's pace, you'll spend $XXX by month end."

Activity Log
Every file read, command executed, API call — risk-labeled, timestamped, live.
Captured via Claude Code hooks. Nothing leaves your machine.

Pre-execution Alerts
High-risk actions (rm, delete, format…) flagged BEFORE they run.
Red banner appears on dashboard. The beginning of "stop, not just watch."

Security Score
Your Claude Code environment scored out of 100.
Checks permissions, denied paths, strictMode, hooks.
Shareable. Per-project scoring included.

CLAUDE.md Editor
See the size of your CLAUDE.md. Click to view and edit it directly.
It's re-sent every turn — most people don't realize how much it costs.

Export
Activity logs as CSV or JSON. Full dashboard report as JSON.

Intelligence Tab
47 projects scored by cognitive load.
Stagnation days. Error rate. What deserves your attention next.

AI Assist Hints
Click any project → get an AI-generated suggestion on how to unblock it.
On-demand. Cached. Uses claude-sonnet-4-20250514 locally.

---

The reaction

Zhangcan Ding (@jason_dingzc) made an explainer video before launch:

"what your AI spent" is table stakes.

"what deserves your attention" is the real insight.

That framing is sharper than anything I wrote myself.

temp.md (@ship_temp_md) said:

turning token spend into what deserves attention next

is a much better loop than just another usage chart.

Both nailed it.

---

Where this is going

This started as a cost visibility tool.

It's becoming something bigger:
the audit layer AI agents have been missing.

Token Visibility → Agent Activity Log → Agent Security → Agent Control

The data was always on your machine.
I'm just making it visible — and eventually, actionable.

npm install -g @notenkidev/claude-token-dashboard
claude-token-dashboard

GitHub: https://github.com/notenkitoclient-cpu/claude-token-dashboard

---

Would love to hear what you're seeing in your own data.