[BUG] Free ultrareview quota consumed when run crashes before producing findings (server-side rate limit)

Resolved 💬 3 comments Opened May 5, 2026 by francobellodi Closed May 9, 2026

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?

A free /ultrareview run launched against Nuttria-Dev/project-red#5 crashed mid-execution due to an Anthropic-side rate limit and produced no findings. The
findings payload returned to the client was just an error stub. Despite zero output being delivered, the run was counted against my free-tier quota: the
counter went from "Free ultrareview 1 of 3" before the crash to "2 of 3" after — i.e., I was charged a free run for an execution the service itself failed
to complete.

What Should Happen?

Ultrareview runs that crash before producing any findings should not consume free-tier quota — especially when the crash is caused by transient
server-side issues (rate limiting, infrastructure failures) rather than anything the user did. The expected behavior is either:

  • Automatic refund of the free run when a crash with no findings is detected, or
  • Exclusion of no-output / crashed sessions from the quota counter.

Error Messages/Logs

Findings payload returned to the client:
  {"error":"Review crashed before producing findings. See session logs for details."}

  Error surfaced when inspecting the session:
  API Error: Server is temporarily limiting requests (not your usage limit) · Rate limited

Steps to Reproduce

Not deterministically reproducible — depends on Anthropic backend rate limiting being triggered during an ultrareview run.

  1. From Claude Code, run /ultrareview <PR_number> against a GitHub PR.
  2. If the backend emits the rate-limit error mid-execution, the session crashes and returns the error stub above.
  3. The free-tier counter still decrements (verified by relaunching: counter went from "1 of 3" → "2 of 3" across the crashed run + successful relaunch).

Claude Model

Opus

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

2.1.128 (Claude Code)

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

Cursor

Additional Information

  • Repo: Nuttria-Dev/project-red (private)
  • PR: #5 — branch feat/1-2-terraform-baseline-aws
  • Date: 2026-05-05
  • Crashed session ID: session_01NFe2KkA88mCxShUnbmcPq4
  • Successful relaunch session ID (no quota issues, same PR, no PR-side changes that would explain the original failure): session_01A5a17AruK5RkM4gB63psim
  • Counter state: "1 of 3" before crashed run → "2 of 3" after, even though crashed run produced no findings
  • The successful relaunch confirms the original crash was infra-related, not caused by anything in the PR contents.

View original on GitHub ↗

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