Issue Station
Sign In
Tools hereToolbox

claude-code - 💡(How to fix) Fix [DOCS] Image input docs missing token-budget/compression details for pasted and attached images [1 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
anthropics/claude-code#45477•Fetched 2026-04-09 08:04:31
View on GitHub
Comments
0
Participants
1
Timeline
2
Reactions
0
Author
coygeekcoygeek
Participants
coygeekcoygeek
Timeline (top)
labeled ×2

Root Cause

That matters because the current docs present paste, drag-and-drop, file attachment, and path-based image input as separate ways to add images, but they do not explain that these paths now converge on the same effective image budget.

RAW_BUFFERClick to expand / collapse

Documentation Type

Missing documentation (feature not documented)

Documentation Location

https://code.claude.com/docs/en/common-workflows

Section/Topic

Work with images, plus the related image-input guidance in desktop, desktop-quickstart, and the Read tool docs.

Current Documentation

The docs explain how to provide images, but they do not explain how pasted/attached images are processed or that they now share the same token budget as images read via the Read tool.

common-workflows currently says:

Suppose you need to work with images in your codebase, and you want Claude's help analyzing image content.

  1. Drag and drop an image into the Claude Code window
  2. Copy an image and paste it into the CLI with ctrl+v (Do not use cmd+v)
  3. Provide an image path to Claude. E.g., "Analyze this image: /path/to/your/image.png"

desktop currently says:

Attach files: attach images, PDFs, and other files to your prompt using the attachment button, or drag and drop files directly into the prompt.

agent-sdk/typescript currently says:

Reads files from the local filesystem, including text, images, PDFs, and Jupyter notebooks.

None of these pages describe image compression/downscaling behavior, token-budget limits for image inputs, or that pasted and attached images are now normalized to the same budget as images provided through the Read tool.

What's Wrong or Missing?

Claude Code v2.1.97 changed user-visible image behavior: pasted and attached images are now compressed to the same token budget as images read via the Read tool.

That matters because the current docs present paste, drag-and-drop, file attachment, and path-based image input as separate ways to add images, but they do not explain that these paths now converge on the same effective image budget.

Users working with screenshots, diagrams, or dense UI images cannot tell from the docs:

  1. that pasted/attached images may be compressed before analysis,
  2. that the Read tool path is now aligned with pasted/attached image handling, or
  3. what consistency to expect when switching between these input methods.

Suggested Improvement

Add a short Image size and processing note to the main image-input docs, ideally in common-workflows and/or desktop, covering:

  • pasted, dragged, and attached images are compressed to fit Claude Code's image token budget,
  • images supplied via the Read tool are handled to the same budget,
  • users should expect broadly consistent analysis across these input methods,
  • large images may be resized/compressed before they are sent for analysis.

If there is a user-relevant size/fidelity caveat (for example, very large screenshots losing detail), document that explicitly where image workflows are introduced.

Impact

Medium - Makes feature difficult to understand

Additional Context

Affected Pages:

PageContext
https://code.claude.com/docs/en/common-workflowsPrimary Work with images workflow for drag/drop, paste, and image-path input
https://code.claude.com/docs/en/desktopAdd files and context to prompts explains file attachments for images
https://code.claude.com/docs/en/desktop-quickstartQuickstart repeats the image attachment and drag/drop guidance
https://code.claude.com/docs/en/interactive-modeDocuments the clipboard shortcut that inserts pasted images
https://code.claude.com/docs/en/best-practicesRecommends pasting or dragging images directly into the prompt
https://code.claude.com/docs/en/agent-sdk/typescriptDocuments that the Read tool accepts image files, which is the comparison point named in the changelog

Total scope: 6 pages affected

Source: Changelog v2.1.97

Exact changelog entry: Improved image handling: pasted and attached images are now compressed to the same token budget as images read via the Read tool

extent analysis

TL;DR

Update the documentation in common-workflows and related pages to include information about image compression, token budget, and consistent analysis across different input methods.

Guidance

  • Add a note to the common-workflows page explaining that pasted, dragged, and attached images are compressed to fit Claude Code's image token budget.
  • Clarify that images supplied via the Read tool are handled to the same budget, ensuring consistent analysis across input methods.
  • Consider documenting any user-relevant size or fidelity caveats, such as large screenshots losing detail, where image workflows are introduced.
  • Review and update the affected pages (desktop, desktop-quickstart, interactive-mode, best-practices, and agent-sdk/typescript) to ensure consistent guidance on image input and processing.

Example

A possible addition to the common-workflows page could be:

Note: Pasted, dragged, and attached images are compressed to fit Claude Code's image token budget. Images supplied via the Read tool are handled to the same budget, ensuring consistent analysis across input methods.

Notes

The suggested improvement should be applied to all affected pages to ensure consistent documentation and avoid user confusion. The exact wording and placement of the added note may vary depending on the page content and style.

Recommendation

Apply the suggested workaround by updating the documentation to include the missing information about image processing and token budget. This will help users understand the changes introduced in Claude Code v2.1.97 and ensure consistent analysis across different input methods.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…
#installation#tensor shape#autograd error#model save/load#optimization

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING