[Feature Request] Add transcript filtering mode to suppress tool output rendering while preserving agent messages

Bug Description
Claude Code currently renders every tool call inline — bash commands, file diffs, read previews, etc. I trust the agent to execute correctly and only read its prose messages; the diffs are visual noise I'm not looking at. On multi-step refactors the transcript is ~90% output I don't engage with. Requested: a settings.json key (e.g. "transcriptMode": "focus" or "hideToolOutput": true) that suppresses the rendered tool activity while keeping the agent's text messages. Tools still run; only the display is filtered. Users who want the full audit can leave it off. Signal this is a real gap: when I asked the agent how to do this, it confabulated a viewMode: "focus" setting (and later "verified" it by hallucinating a docs line), and I only caught the mistake because nothing changed after restart. A first-class toggle would also stop this class of hallucination — the agent would have a real setting to point at.

Environment Info

• Platform: linux
• Terminal: windows-terminal
• Version: 2.1.114
• Feedback ID: a427ae51-9c6b-4a35-b924-a16b4ce8d484

Errors

[{"error":"MaxFileReadTokenExceededError: File content (25606 tokens) exceeds maximum allowed tokens (25000). Use offset and limit parameters to read specific portions of the file, or search for specific content instead of reading the whole file.\n    at uD7 (/$bunfs/root/src/entrypoints/cli.js:4725:12550)\n    at processTicksAndRejections (native:7:39)","timestamp":"2026-04-18T17:34:15.857Z"}]