/ultrareview crashed twice on large PR — consumed 2/3 free credits without producing findings

Resolved 💬 3 comments Opened May 5, 2026 by michaelchevallier Closed May 5, 2026

Summary

/ultrareview <PR#> crashed twice on a large PR, consuming 2 of 3 free ultrareview credits without producing any findings. Requesting credit restoration.

Sessions

Both returned: {"error":"Review crashed before producing findings. See session logs for details."}

PR details

  • Repo: michaelchevallier/arbitrage (private)
  • PR: #2 — 69 commits, 221 files changed, ~25k insertions / ~3.3k deletions
  • Base branch: phase7-base, head: review/phase7

Steps to reproduce

/ultrareview 2
# → "Free ultrareview 1 of 3. Ultrareview launched..."
# → task-notification: {"error":"Review crashed before producing findings."}

/ultrareview 2
# → "Free ultrareview 2 of 3. Ultrareview launched..."
# → task-notification: {"error":"Review crashed before producing findings."}

Expected behavior

Either the review completes and produces findings, or if the PR is too large, the service should reject it before consuming a credit.

Request

Please restore the 2 consumed credits. The crashes appear to be caused by the PR size — if there's a size limit, an upfront validation error would prevent this.

View original on GitHub ↗

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