Aimprint — Your AI usage footprint

Track and understand your Claude Code token consumption across machines, projects, and types of work.

Live dashboard → danforthhh.github.io/Aimprint

---

What it does

Claude Code's /cost only tells you "you're using your subscription." Aimprint gives you the full picture:

• How many tokens you consumed, by day / project / model / machine
• What type of work drove your usage (code writing, deep analysis, refactoring, planning…)
• This week vs last week — period-over-period comparison on every view
• Sub-agent breakdown — what % of your tokens came from direct use vs Claude's background agents
• Ticket-level tracking — tokens per feature branch / Jira ticket
• All your machines in one view — sync from as many laptops as you want

Everything runs on free infrastructure. The tracker itself costs zero tokens to operate.

---

Architecture

~/.claude/projects/*.jsonl          # Claude Code session logs (local)
        │
        ▼
  sync agent (Node.js)              # runs on each machine, classifies sessions
        │  POST /ingest
        ▼
  Cloudflare Worker + D1            # multi-tenant API + SQLite storage (free tier)
        │  GET /api/*
        ▼
  React dashboard (GitHub Pages)    # charts, filters, session table
        │
  Firebase Auth                     # email/password login, per-user data isolation

---

Quick start

1. Register

Go to danforthhh.github.io/Aimprint → create an account.

2. Get a sync token

Settings → Generate Sync Token → copy it (shown once).

3. Install the sync agent

git clone https://github.com/Danforthhh/Aimprint
cd Aimprint
npm install
cp sync/.env.example sync/.env

Edit sync/.env:

WORKER_URL=https://aimprint.vin-bories.workers.dev
SYNC_TOKEN=<your-token>

4. Run your first sync

npm run sync

5. Automate (optional)

Add to ~/.claude/settings.json to sync at every session start:

{
  "hooks": {
    "SessionStart": [{
      "hooks": [{
        "type": "command",
        "command": "npm run sync --prefix /path/to/Aimprint >> ~/.claude/aimprint-sync.log 2>&1 &",
        "timeout": 30
      }]
    }]
  }
}

Multiple machines

Use the same sync token on every machine. All machines merge under your account and appear separately in the "machine" filter.

---

Session categories

Sessions are classified locally by the sync agent — no external API calls, zero token cost.

Category	What it means
Code writing	Implementing features, writing new code
Code process	CI/CD, deployments, Docker, builds
Quality	Tests, linting, code review, security
Deep analysis	Codebase exploration, architecture, research
Refinement	Refactoring, simplifying, cleaning up
Planning	PRDs, tickets, strategy, roadmaps
Random	Quick questions, short sessions
Other	Doesn't fit a clear pattern

You can manually re-assign any session's category from the session table.

---

Tech stack

Layer	Tech
Frontend	React 19 + TypeScript + Vite + Tailwind CSS 4 + Recharts
Auth	Firebase Auth (email/password)
API + storage	Cloudflare Worker + D1 (SQLite) — free tier
Sync agent	Node.js + TypeScript (tsx)
Deployment	GitHub Pages (npm run deploy)

---

Self-hosting

Want to run your own instance? See docs/obsidian/Setup/Deploy.md for the full setup guide (Cloudflare + Firebase + GitHub Pages, all free).

---

Privacy

• Conversation content never leaves your machine — only token counts and metadata are synced
• First message text (up to 500 chars) is used locally for session classification and stored in the cloud under your account only
• No third-party analytics, no ads