Add tunable verbose output showing Claude's background activity#29
Merged
RichardAtCT merged 6 commits intomainfrom Feb 18, 2026
Merged
Add tunable verbose output showing Claude's background activity#29RichardAtCT merged 6 commits intomainfrom
RichardAtCT merged 6 commits intomainfrom
Conversation
Introduces a VERBOSE_LEVEL setting (0-2) and /verbose command that lets users control how much of Claude Code's background work is shown in real-time via Telegram progress messages: - Level 0 (quiet): only "Working..." and final response - Level 1 (normal, default): shows tool names as they execute - Level 2 (detailed): shows tool names + input summaries (filenames, commands, search patterns) The progress message is updated every 2 seconds during execution with a running list of tool calls. The stream callback is wired into all three agentic handlers (text, document, photo). SDK streaming now also extracts ToolUseBlock data for the callback. https://claude.ai/code/session_01RZGrZTQsvypQQM2EP5F24v
Telegram's typing indicator expires after ~5 seconds. Now all three agentic handlers (text, document, photo) re-send the typing action every 2 seconds via the stream callback, so the user always sees the bot is working. This works at all verbose levels including 0. https://claude.ai/code/session_01RZGrZTQsvypQQM2EP5F24v
The stream callback now captures assistant text content alongside tool
calls. At verbose level 1, the first line of each text block is shown
(truncated to 80 chars). At level 2, up to 120 chars are shown. This
lets users see Claude's reasoning ("I'll check the test file first",
"Let me try a different approach", etc.) interleaved with tool activity
in the progress message.
https://claude.ai/code/session_01RZGrZTQsvypQQM2EP5F24v
The verbose output now shows Claude's intermediate reasoning text alongside tool calls. Updated CLAUDE.md, settings field description, and /verbose help text to accurately describe all three levels. https://claude.ai/code/session_01RZGrZTQsvypQQM2EP5F24v
Add /verbose to the agentic mode commands list, update the demo to show real-time progress output, add a verbosity levels table, add VERBOSE_LEVEL to the configuration section, and note the features in the working features list. https://claude.ai/code/session_01RZGrZTQsvypQQM2EP5F24v
|
PR Review Summary
What looks good
Issues / questions
Suggested tests (if needed)
Verdict
|
Two security/reliability fixes for the verbose output system: 1. Secret redaction: _summarize_tool_input now runs Bash command previews through _redact_secrets() which strips API keys, tokens, passwords, AWS keys, Bearer/Basic auth, and connection string credentials before they reach Telegram chat. 2. Typing heartbeat: Replace stream-event-coupled typing indicators with an independent asyncio background task that sends "typing" every ~2s regardless of stream cadence. This keeps the indicator alive during long tool executions with sparse/no intermediate output. The heartbeat is cancelled in a finally block. Includes 15 new tests covering both features. https://claude.ai/code/session_01RZGrZTQsvypQQM2EP5F24v
This was referenced Feb 18, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Introduces a VERBOSE_LEVEL setting (0-2) and /verbose command that lets
users control how much of Claude Code's background work is shown in
real-time via Telegram progress messages:
commands, search patterns)
The progress message is updated every 2 seconds during execution with a
running list of tool calls. The stream callback is wired into all three
agentic handlers (text, document, photo). SDK streaming now also extracts
ToolUseBlock data for the callback.
https://claude.ai/code/session_01RZGrZTQsvypQQM2EP5F24v