[BUG] Ultrareview: failed runs reported as completed to agent

Preflight Checklist

• [x] I have searched existing issues and this hasn't been reported yet
• [x] This is a single bug report (please file separate reports for different bugs)
• [x] I am using the latest version of Claude Code

What's Wrong?

I ran /ultrareview twice on PR #1073 of bennycode/trading-signals and both runs failed:

<img width="796" height="428" alt="Image" src="https://github.com/user-attachments/assets/8d51e61e-7672-48b7-be7d-8d1a35109d28" />

Unfortunately, both runs sent a task-notification to the agent with status=completed
and findings=[]. The agent interpreted that as a clean review with no
findings. When I opened the session URLs in the web UI, however, both
show "Review failed".

What Should Happen?

Either:

• The run succeeds and the UI + agent both reflect that, OR
• The run fails and the agent gets a status=failed notification AND the failed run does not consume my free quota.

Error Messages/Logs

Steps to Reproduce

1. Use my open PR: https://github.com/bennycode/trading-signals/pull/1073
2. Launch /ultrareview from Claude CLI, target feat/trading212-exchange to main

Claude Model

Opus

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

2.1.138

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Terminal.app (macOS)

Additional Information

_No response_