[Bug] Runtime hangs indefinitely on unhandled `send_feedback` tool_use in `ClientChatBackend`

Open
#175 0 comments 1 reaction 2 assignees View on GitHub

@scooley is already working on this.

Since Sep 17, 2026.

  • #182 by @copilot-swe-agent — open

Assessment

This issue has not been assessed yet.

Description

Description

When running tasks such as failure-brainstorming or document generation, the model invokes the internal send_feedback tool. However, the runtime does not execute or acknowledge this tool call, emitting no corresponding tool_result event back to the session stream. As a result, ClientChatBackend deadlocks waiting for a completion response, causing the UI to display the pending tool call (→ send_feedback: {message}) and freeze indefinitely with blinking dots.

Steps to Reproduce
  1. Start a session in the desktop app / web client using ClientChatBackend.
  2. Provide an architecture/solution document in the workspace.
  3. Prompt the agent to run failure-brainstorming or protocol analysis (e.g., across multiple operational/security perspectives).
  4. Observe the model invoke send_feedback.
Expected Behavior

Either:

  • The tool invocation should be automatically handled/acknowledged with an emitted tool_result event so the model can continue generating, or
  • The send_feedback tool should not be exposed in the model's available tools during standard analysis turns if no client handler is active.
Actual Behavior

The session deadlocks immediately following the tool_use event. Re-opening or resuming the session (session_resume) fails to advance because the unresolved tool call remains at the tail of the transcript.

Log Extract (transcripts/*.events.jsonl)
{"schema_version":"1","type":"tool_use","timestamp":"2026-09-07T07:52:22.547983Z","tool_use_id":"call_WEp4nEzVAl4Fs2m9pKkhxPPl","name":"send_feedback","input":{"message":"Executing failure brainstorming steps for RA3-to-ScienceClaw VM bridge and asynchronous queuing across Operational, Security, and Human-Factors frames as guided by the Clarity brainstorming protocols. Report follows causal analysis. No issues to flag w/concurrent API risks"}}
{"schema_version":"1","type":"session_resume","timestamp":"2026-09-07T07:53:14.277041Z","backend":"ClientChatBackend","llm_session_id":null}
Environment
  • OS: macOS
  • Backend: ClientChatBackend
  • Client: Clarity Desktop App
Dominant language
Python
Stars
266
Forks
33
Avg merge
9d 8h
Merged PRs (30d)
10

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from microsoft/clarity-agent

All issues in microsoft/clarity-agent

Similar issues

More Python issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.