read-large-md

name	read-large-md
description	Read markdown files that are too large for the Read tool due to embedded base64 images. Use when Read fails with "exceeds maximum allowed size" on a .md file.
user-invocable	false

Reading Large Markdown Files with Embedded Images

Markdown files exported from Google Docs or similar tools often contain base64-encoded inline images (![alt](data:image/png;base64,...)) on single very long lines. These make the file hundreds of KB or millions of tokens, causing the Read tool to fail with "exceeds maximum allowed size" or "exceeds maximum allowed tokens."

Technique

Use Grep to extract only the text lines, filtering out image data:

Grep pattern="^[^!]" path="<file>" output_mode="content" head_limit=200

This matches all lines that do NOT start with ! (markdown image syntax), skipping the ![...](data:image/png;base64,...) lines.

Paginating through results

Use offset and head_limit to read in chunks:

Grep pattern="^[^!]" path="<file>" output_mode="content" offset=0 head_limit=200
Grep pattern="^[^!]" path="<file>" output_mode="content" offset=200 head_limit=200

Finding specific sections

Search directly for section headers or keywords:

Grep pattern="## Theory of Change" path="<file>" output_mode="content" -A=30

When to use this

• Read tool fails on a .md file due to size
• The file is known to contain embedded images (e.g., Google Docs export)
• You need the text content but not the images