Loading page content
Claude re-reads the whole conversation on every message, so the 40th message in a chat can cost ten or twenty times what the first one did. Describe how you work and this planner shows where your usage goes and when a fresh chat would save it.
Every Claude message re-reads the entire conversation, so a long chat costs far more than the same questions spread across a few short ones. Describe how you work and see where your usage actually goes, and when to start a new chat. Runs in your browser; nothing is sent anywhere.
Before I start a fresh chat, write a handoff note I can paste into it. Keep it under 300 words and include: 1. The goal and the current status in two sentences. 2. Decisions we made and why (bullets). 3. Facts, numbers, names, and file names the next chat must not get wrong. 4. What is still open, and the very next step. Do not include pleasantries or a recap of things we discarded.
Anthropic doesn't publish limits in tokens, so this shows relative usage, not a countdown. Token counts are estimates (~1.33 tokens per word, ~500 per page), and Project caching, tool use, effort level, and model choice shift the absolute numbers. The shape of the curve holds either way. Full guide: why you keep hitting Claude's usage limit. Related: Claude Code Context Auditor, Token Counter.
100 of 102 run entirely in your browser (Weather Widget uses Open-Meteo; What's My IP asks this site's server).
Share on XShare on LinkedInThe most common reason people hit Claude's limit early is not asking too many questions. It's asking them all in one chat. Each new message carries the full history with it, including every attachment and every earlier reply, so usage per message climbs in a straight line and the chat's total climbs with the square of its length. A 60-page PDF in a 25-message chat is read 25 times.
Anthropic doesn't publish the limits in tokens, and they vary by plan, model, and time of day, so this tool shows relative usage rather than a countdown. That's the useful part anyway: the same work, structured differently, can use a fraction of the allowance. The planner picks the longest chat length that captures most of the savings, because splitting too often loses context that the handoff note can't carry.