DEV Community: Deibyg

I spent $514 on Claude Code in 30 days. Here's what I learned."

Deibyg — Sun, 10 May 2026 21:25:33 +0000

I spent $514 on Claude Code in 30 days. Here's what I learned.

Real data from 50 sessions. The good, the bad, and the expensive.

The setup

For the past month, I've been tracking every Claude Code session with claudestat — an open-source tool I built to monitor token usage, costs, and patterns.

This isn't a hypothetical study. These are my real numbers from real work on real projects.

The numbers

30-day period: April 12, 2026 → May 10, 2026

Metric	Value
Total spent	$514.86
Sessions	50
Total tokens	3,803,319
Loops detected	375
Avg efficiency	69/100

That's about $18/day on average. For context: I'm on Max 5, so this isn't cheap.

The most expensive session

April 26, 2026 — I let Claude run on a refactoring project for hours without checking:

Cost: $32.94
Loops detected: 31
Efficiency score: 35/100
What happened: Claude went in circles editing files, each edit triggering more edits. I wasn't watching.

That's $33 in one session. More than my daily coffee budget for a month.

Key lesson: Loops are expensive. 31 loops × average tool call cost = $33 down the drain.

The project breakdown

Who consumed the most?

Project	Spent
claudetrace (side project)	$326.44
wodrival	$61.32
claudestat	$60.35
conductor	$39.40
Other	$27.85

63% of my spending went to one side project. It's not my main job — it's my hobby project that's burning my quota.

This is the kind of insight you only get from tracking.

The loop problem

375 loops detected across 50 sessions.

That's ~7.5 loops per session on average. But it's skewed:

12 sessions had more than 10 loops
9 sessions had efficiency below 50%

Most loops look like:

Read → Edit → Edit → Edit → Read → Edit → Edit → Edit

Claude tries something, it doesn't work, tries slightly different, repeats. This is where most of my waste came from.

What I learned

1. Loops are the silent killer

I wasn't aware of loops until I saw the data. Now I watch for:

Same tool called 3+ times in a row
Context creeping up while output stays flat
"Let me try..." messages from Claude

2. One project dominates

63% on claudetrace — I would have guessed 40/60 at best. Tracking revealed the truth.

3. Efficiency varies wildly

Best session: 100/100 (perfect)
Worst session: 35/100 (that $32.94 day)

Understanding this helps me spot when I'm about to have a bad session.

4. Bash is my most expensive tool

Running claudestat top shows Bash at 35-45% of total cost across all sessions. Batch your commands.

How to track your own data

npm install -g @statforge/claudestat
claudestat install
claudestat start
# ... use Claude Code normally ...
claudestat export json > my-data.json

Then analyze:

Total spent → claudestat export | jq '.[].total_cost_usd | add'
Top projects → claudestat export | jq 'group_by(.project_path) | map({project: .[0].project_path, cost: map(.total_cost_usd | add)})'
Problem sessions → Look for efficiency < 50 or loops > 10

What's next

Now that I have this data:

I set up alerts at 70%, 85%, 95% quota → no more surprises
I use the kill switch at 95% → prevents runaway sessions
I check top tools weekly → batch those Bash commands
I watch for loops → stop them early

The black box is open. You have the data.

Try it

npm install -g @statforge/claudestat

Questions? Drop them in the comments.

Data from my actual sessions. Not estimated. Not hypothetical.

Claude Code quota management: a complete guide (2026)

Deibyg — Sun, 10 May 2026 21:18:18 +0000

How to track, monitor, and control your Claude Code usage before it eats your budget.

The problem

You're in the middle of a coding session. Your Claude Code Max 5 limit just hit — right before a critical deadline. You have no idea when it happened, why, or how much you've spent this month.

That's the reality for thousands of developers using Claude Code. The tool is powerful, but it's a black box — you can't see token usage in real time, you're not warned before hitting limits, and there's no built-in way to control spending.

This guide covers everything you need to manage Claude Code quota effectively in 2026.

Understanding Claude Code quota

Claude Code offers three plans:

Plan	Monthly prompts	Best for
Pro	500K tokens/day	Solo developers, experimentation
Max 5	5 hours + 100K tokens/cycle	Heavy daily users
Max 20	20 hours + 300K tokens/cycle	Professional teams

The tricky part: quota resets every 5 hours (for Pro/Max 5) or rolls over (Max 20). It's easy to lose track.

What counts toward quota

Input tokens (everything you type)
Output tokens (everything Claude generates)
Cache read tokens (previously cached context)
Cache write tokens (new context saved)

Key insight: Cache hits don't count fully toward your quota — they're discounted at 10%. This makes efficient context reuse valuable.

How to track your usage in real time

The solution is claudestat — an open-source monitoring tool that hooks directly into Claude Code:

npm install -g @statforge/claudestat
claudestat install
claudestat start
open http://localhost:7337

Now you see:

Live tool trace — every call with duration and estimated cost
Context percentage — see the context approaching the limit before it happens
Burn rate — tokens per minute, projected weekly/monthly spend

This is the "htop for Claude Code" — you finally see what's happening.

Setting up alerts before you hit the wall

claudestat polls your quota every 60 seconds and sends alerts:

# Default: alerts at 70%, 85%, 95%
claudestat config --alerts true

# Or check manually anytime:
claudestat status

Output:

Quota 5h   45/50 prompts (90%)  |  reset in 22m
Plan        MAX5
Burn rate   1,240 tok/min

You'll see a warning before the quota hits — no more surprise "cycle exhausted" messages mid-session.

The kill switch: auto-blocking new sessions

For power users, claudestat has a kill switch — it automatically blocks new Claude Code sessions when you hit a threshold:

# Block new sessions at 95% quota
claudestat config --kill-switch true --threshold 95

This prevents the "I accidentally burned my whole cycle" scenario. No more losing a day's work because Claude kept running.

Analyzing which tools cost the most

You know overall cost. But what if you knew which specific tools are draining your quota?

claudestat top --days 30

Output:

🏆 claudestat top — by est. cost (last 30 days)

#  Tool              Calls    Duration   Est. Cost      %
──  ─────────────────  ────────  ────────────  ─────────  ────
1   Bash              1,240      18.3m      $1.24    38%
2   Read               890       4.1m      $0.87    27%
3   Edit               430       2.8m      $0.61    19%
4   Agent (haiku)      120       9.2m      $0.38    12%
5   Write              210       1.1m      $0.12     4%

Key insight: Bash is typically 35-45% of total cost. You can immediately optimize by batching shell commands.

Loop detection

claudestat also detects inefficiency patterns — like repeated tool calls (loops):

# In the dashboard, see "Loops detected" for each session
# Or in the trace, look for patterns like:
# Read → Edit → Edit → Edit → Read → Edit → Edit

This is often 15-25% of wasted quota — money you didn't need to spend.

Exporting data for reporting

Need to report Claude Code costs to your boss or for expense tracking?

# Export as JSON
claudestat export json --output ~/claude-costs.json

# Export as CSV for spreadsheets
claudestat export csv --output ~/claude-costs.csv

# Filter by date range
claudestat export json --from 2026-04-01 --to 2026-04-30

Each row includes:

id, started_at, cwd, project_path
total_cost_usd
total_input_tokens, total_output_tokens
efficiency_score
loops_detected

Now you can build monthly reports, track trends, or justify Claude Code as a business expense.

Summary: your quota management stack

Task	Solution
Real-time monitoring	`claudestat start` + dashboard
Alert before hitting limit	`claudestat config --alerts true`
Kill switch	`claudestat config --kill-switch true`
Analyze tool costs	`claudestat top`
Export for reporting	`claudestat export`
Health check	`claudestat doctor`

All data stays local — zero cloud dependencies. Your usage data never leaves your machine.

What's next

Claude Code quota management in 2026 is about visibility + control:

Install claudestat — takes 2 minutes
Set up alerts — sleep better knowing you'll be warned
Use the kill switch for critical sessions
Review your top tools weekly — optimize your expensive habits
Export monthly — track your actual spend

The black box era is over. You have the tools to see, control, and optimize your Claude Code usage.

Try it:

npm install -g @statforge/claudestat

Questions? Drop them in the comments or open an issue on GitHub.

I built a real-time monitor for Claude Code after 355 undetected loops burned my quota

Deibyg — Sun, 10 May 2026 17:35:52 +0000

Building dev tools for the AI coding era. Creator of claudestat —
real-time token monitor for Claude Code. TypeScript · Node.js · React.
Always shipping, always learning.

npmjs.com

Claude Code is powerful—but a black box: how much is it spending? looping? how much context is left? I built claudestat: a real-time dashboard with costs, tool calls, loop detection, and reports. npm i -g @deibygs/claudestat full visibility.

Deibyg — Mon, 04 May 2026 01:26:20 +0000