Response silently truncated/hung after reading many images in a single turn

Resolved 💬 3 comments Opened Apr 6, 2026 by xiaodream551-a11y Closed May 16, 2026

Description

When reading 12+ JPEG images using the Read tool within a single conversation turn, the assistant produces no text output after the final tool call and the conversation appears to hang/freeze silently.

Steps to Reproduce

  1. Start a conversation with Claude Opus 4.6 (1M context) in Claude Code CLI
  2. Ask it to analyze content from a MOBI ebook that has been unpacked into individual JPEG page images
  3. The assistant reads 12+ images across multiple parallel Read tool calls to extract formulas and diagrams
  4. After the last batch of Read calls returns successfully, the assistant generates no text response — the turn ends silently with no output

Expected Behavior

The assistant should produce a text response summarizing or analyzing the images it just read, even if context is getting large.

Actual Behavior

The turn completes with zero text output. No error message is shown. The user has to send another message to prompt the assistant to continue.

Environment

  • Model: Claude Opus 4.6 (1M context), max effort
  • Platform: macOS (Darwin 25.4.0), MacBook Air M4
  • CLI version: Claude Code CLI
  • Working directory context: Reading ~12 JPEG images (each ~50-200KB) from an unpacked MOBI textbook

Additional Context

  • The images were read using the built-in Read tool (which supports image files)
  • Each image was a scanned textbook page containing mathematical formulas
  • The conversation had moderate prior context from text extraction and earlier image reads
  • A follow-up message ("Continue from where you left off") successfully got the assistant to produce the intended response, suggesting the content was processable — just silently dropped on the first attempt

🤖 Generated with Claude Code

View original on GitHub ↗

This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗