/ultrareview returns empty findings after rate limit error — burns free trial slots
Resolved 💬 5 comments Opened Apr 24, 2026 by JayIsCoding Closed Jun 8, 2026
Summary
/ultrareview is consuming free trial slots even when the underlying review fails due to a server-side rate limit. Both of my first two free trials returned empty findings ([]) despite a large diff that could not plausibly be clean.
Environment
- Claude Code CLI
- Command:
/ultrareview(no args, reviewing current branch againstmain) - Branch:
kg-ultrareview-2026-04-24→main - Diff scope: 41 files changed, 22,439 insertions(+), 1,687 deletions(-)
Reproduction
- Run
/ultrareviewon a branch with a large diff vs.main. - Observe "Free ultrareview X of 3" counter decrement.
- Wait ~5–10 min for the
task-notificationto arrive. - Findings payload returns as
[].
Observed Behavior
- Run 1:
session_01FwYeGQ15mhZrthPqycLVdf— explicit "Server is temporarily limiting requests" / rate-limited error observed, task still markedcompleted, findings returned as[], free-trial counter decremented from 3 → 2. - Run 2:
session_01LugWVBs5zwqdTvDsLK29qt— same outcome, counter decremented from 2 → 1. - No reviewer agent output, no diff coverage summary, no partial findings.
Expected Behavior
- A real review of a 22k-line diff across 41 files should surface something (style, correctness, security, or at minimum scope notes).
- When the platform rate-limits the reviewer ensemble and the review cannot complete, the run should not decrement the free-trial counter.
- Failed/empty runs should be surfaced to the user as errors, not as
completedwith[]findings.
Ask
- Refund the 2 consumed trial slots on this account so I can actually use the feature.
- Do not decrement the free-trial counter when
/ultrareviewfails due to a server-side error (rate limit, reviewer crash, empty payload). - Distinguish "review completed, nothing found" from "review failed to run" in the task-notification — currently both look identical to the user.
Evidence
- Session 1 tracking URL: https://claude.ai/code/session_01FwYeGQ15mhZrthPqycLVdf
- Session 2 tracking URL: https://claude.ai/code/session_01LugWVBs5zwqdTvDsLK29qt
- Task IDs:
rgr6zuqzp,rrn7fw9dc
This issue has 5 comments on GitHub. Read the full discussion on GitHub ↗