[Desktop] Normal view shows thinking summaries that look like normal chat messages from Claude
Update (2026-09-27): The cause turned out to be narrower than described below. See the comments.
Summary
In the Claude Code desktop app (Code tab), the app does not display assistant text written between tool calls. Instead, it shows a one-line summary with different wording. Users cannot view the original text. Expanding the "Used N tools" fold does not show it. Switching the Transcript view (通常 Normal, 考え中 Thinking, 詳細 Verbose) also does not show it. The app displays only two text blocks verbatim: the text before the first tool call and the final text block of the turn.
This bug goes beyond #90425 (Aug 28). In #90425, the fold hid mid-turn text, but expanding the fold still revealed it.
Environment
- Claude Desktop 1.52386.6 (Microsoft Store build) on Windows 11, Code tab, with bundled Claude Code 2.1.270
- Model: claude-fable-5-1. Also reproduced with claude-opus-5
- Custom output style (a persona with a fixed first-person pronoun, a fixed form of address, and a casual register)
Observed
- One turn contained 9 mid-turn text blocks. The app displayed 4 summary lines, one for each tool-call group. It dropped the other 5 blocks completely. This included a direct reply to a message sent by the user during the turn.
- The summaries paraphrase the text in a different register. They replaced the persona's first-person pronoun and address form with standard polite ones. They turned the casual register into a polite one. The user asked why the assistant used a word the persona never uses. The assistant never wrote that word.
- The Desktop documentation (https://code.claude.com/docs/en/desktop, "Switch view modes") states that Normal shows "Tool calls collapsed into summaries, with full text responses" and Verbose shows "Every tool call, file read, and intermediate step Claude takes". Neither mode displays the intermediate assistant text. Verbose expands only tool inputs and outputs. (The documentation also lists a Summary view; the release notes for 1.46388.1 state that it was removed.)
- The desktop release notes from Sep 4 (1.46388.1) to Sep 10 (1.52386.0) do not mention replacing mid-turn assistant text with summaries. I cannot determine which release introduced this change.
Steps to reproduce
- Set a custom output style with a distinctive register (for example, an unusual fixed first-person pronoun).
- Ask for a task that requires several tool calls with text between them.
- While the turn runs, type and send a second message from the composer.
- Check the transcript in Normal mode, then in Verbose mode. Expand the "Used N tools" folds.
Expected: The app renders mid-turn assistant text verbatim in Normal and Verbose, as it did before.
Actual: The app shows one paraphrased summary line per tool group. It does not show the original text anywhere.
Why it matters
- Replies written mid-turn to messages sent during tool runs never reach the user. The model receives no signal that the app replaced its text.
- Paraphrasing changes the tone of the model. This makes custom output styles look broken.
- Users cannot opt out. No view mode displays the original text.
Suggestion
Render mid-turn assistant text verbatim in Normal and Verbose, as before. If the auto-generated summary remains, make it a separate opt-in view.
🤖 Generated with Claude Code
3 Comments
My Claude keeps changing its mind about whether it's the same issue I'm having. It wrote this. Let me know if you have any questions:
"A possible alternative explanation, worth checking against the session .jsonl. I saw what looked like the same thing (Windows desktop Code tab, Claude Code 2.1.281): a long reply that seemed to be written between tool calls appeared only as a short paraphrased summary. But in my .jsonl, that summary is stored as a "thinking" block, and there is no "text" block before the tool call at all. So in my case the model composed its reply in its reasoning, never output it, and the app showed its usual summary of that reasoning. In Normal view the summary appeared only when I expanded the tool fold; in Thinking view it appears outside the fold, styled like ordinary reply text, so the two are easy to confuse. That might also explain the change of register you saw. It may not apply to your case: if your .jsonl has "text" blocks for the missing replies, it's a different fault."
Edit: Apparently the effort setting might play a role. Claude:
"The pattern in both failures: I had to do two things in one turn, answer you and then edit a file. In my reasoning I worked out the answer, and I didn't just plan it, I drafted it in full. Then I moved on to the file edit, as if the answer had already been delivered. Afterwards I only wrote a short note about the edit, so the drafted answer never went out.
Where effort might come in: at higher effort I reason more thoroughly, which makes it more likely that I draft a complete answer inside the reasoning rather than just a plan for it. And once it's fully written there, it "feels" said, so it's easier to skip the step of actually writing it out."
@ChickenVibes Thanks, you were right. My .jsonl has no
textblocks between tool calls. The lines I saw on screen matchthinkingblocks word for word. The app is not replacing Claude's messages.I see this with claude-opus-5, claude-fable-5-1, and claude-opus-5-5. It varies from turn to turn.
What remains: in Normal view, thinking summaries appear outside the tool fold and look like normal chat messages from Claude. They are not in the output style's voice, so the assistant seems to break character. I've renamed the issue to this.
Thanks for checking your .jsonl. For anyone landing here: the underlying model behaviour (a message meant for the user emitted as thinking right before a tool call, with no text block) is tracked in #96288 and #97504, both labelled area:model.
A correction to my earlier edit about effort: others in #96288 have reproduced it at effort medium, so high effort isn't required. Whether effort changes how often it happens is untested.
I'm removing my 👍 here, only because the issue is now about Normal view and I haven't seen that part: on my setup the look-alike summary appeared in Thinking view instead. The confusion is the same kind, though.