---
title: "Page token budget · Sitebulb Labs"
description: "A page that does not fit an agent's context window gets truncated, so some of its content may be missed."
url: https://labs.sitebulb.com/docs/checks/agent-experience/page-token-budget/
---

# Page token budget

A page that does not fit an agent's context window gets truncated, so some of its content may be missed.

- Category

  [Agent experience](https://labs.sitebulb.com/docs/checks/agent-experience)

- Standard

  Recommended

## What it checks

The extension counts the characters of text on the page. It uses the rendered page when it can read it, and falls back to the text in the server’s HTML when it cannot. The check passes at **100,000 characters or fewer**, which is roughly 25,000 tokens at four characters per token.

Past that point, an agent working within a context limit is likely to truncate the page, and whatever falls after the cut-off never gets read.

## Results

| Status         | When                                                                  |
| -------------- | --------------------------------------------------------------------- |
| **Pass**       | The page has 100,000 characters of text or fewer                      |
| **Warn**       | The page has more than 100,000 characters of text                     |
| **Unmeasured** | The rendered page could not be read and the HTML could not be fetched |

The result shows the character count and an approximate token count either way.

## How to fix

Split oversized pages into focused documents, one topic each, and link them from an index page. Long reference pages, full changelogs and single-page documentation are the usual cases.

If you publish an [llms.txt](https://labs.sitebulb.com/docs/checks/content-accessibility/llms-txt), list the smaller pages there so an agent can fetch only the one it needs.