[BUG] Cowork: "No conversation found with session ID" on every new task (Windows 11)

Resolved 💬 3 comments Opened May 7, 2026 by kimseokho1226 Closed May 11, 2026

Preflight Checklist

  • [x] I have searched existing issues and this hasn't been reported yet for the Cowork desktop tab
  • [x] This is a single bug report
  • [x] I am using the latest version of Claude Desktop

What's Wrong?

Cowork (beta) on Claude Desktop consistently fails with "No conversation found with session ID: <uuid>" on Windows 11. The error appears almost immediately after the task starts, making the feature unusable.

The Retry button does not recover the session — it produces the same error with the same session ID. New tasks fail the same way.

This looks like the same family of regressions tracked in #36865 / #36842 / #57102, but I'm filing separately because:

  • This is the Cowork tab in Claude Desktop, not VS Code teleport.
  • It reproduces on every new task, not only on resume.

Steps to Reproduce

  1. Open Claude Desktop on Windows 11 (Max plan).
  2. Switch to the Cowork tab.
  3. Create a New task (project context: a custom project, but reproduces without a project too).
  4. Send any short prompt.
  5. Task runs briefly, then errors out with:

No conversation found with session ID: c9c7bce1-cb67-4539-af63-287ef6a1ea12

Expected Behavior

The task should accept the message and continue execution against the assigned session.

Actual Behavior

  • The Cowork session errors out with No conversation found with session ID: <uuid>.
  • "Retry" reproduces the same error with the same session ID.
  • Creating a new task produces the same error with a fresh session ID.
  • The progress sidebar shows the task partially advanced (2 of 3 steps checked) before the failure.

Screenshot: _(drag your screenshot into the issue editor here before submitting)_

Environment

  • OS: Windows 11
  • App: Claude Desktop (latest, Cowork beta enabled)
  • Plan: Max
  • Locale: ko-KR
  • Project context: enabled (custom project)
  • MCP servers configured: 20+ stdio servers (desktop-commander, ssh-synology, naver-search, law-info-v2, nh-coop, saju, jamidusu, etc.) — none are invoked before the failure.
  • Model selected: Claude Opus 4.7

What I've Tried

  • Pressing Retry → same error, same session ID.
  • Creating a new task (same project) → new session ID, same error.
  • Creating a new task without project context → same error.
  • Restarting Claude Desktop fully → no change.
  • Local Chat / Code tabs in the same Desktop app → working normally.

Additional Context

  • Anthropic Status posted a related infrastructure incident on 2026-05-06 → resolved 2026-05-07 12:45 UTC (GitHub outbound IP change affecting Claude Code remote sessions). Cowork-side errors persist into 2026-05-08 KST, so the symptom likely outlasted that incident or is a separate Cowork regression.
  • Similar pattern reported in:
  • #36865 (VS Code, Windows — every new cloud session fails immediately)
  • #36842 (VS Code, Windows — teleport returns 0 log entries)
  • #57102 (area:cowork, Windows, regression — opened 2026-05-07)

Happy to provide additional logs from %APPDATA%\Claude\logs if useful — let me know which files are relevant.

View original on GitHub ↗

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