ultrareview: failed Find phase consumed free quota slot (empty results, ~10s heartbeats)

Summary

A /ultrareview invocation completed in a "Review failed" state with empty Find output, but my free quota was decremented. Requesting the slot be credited back.

Session

• Session URL: https://claude.ai/code/session_01NHxadrPFJHGxtcXaWZi9Eg
• Repo: private repo (majinjacques/btcbot), Claude GitHub app installed and confirmed access
• Branch: review-all → base main (initial commit, ~297K-line diff against empty base)
• Date: 2026-04-25

What happened

UI showed:

• ✓ Set up cloud container
• ✓ Cloned repository
• ○ Run setup script — not run (no setup.sh existed yet)
• ✓ Started Claude Code

Then Review failed:

• ✓ Setup
• ✗ Find — output was just two heartbeats then an empty array:

```
[heartbeat] 10:42:52
[heartbeat] 10:43:02

<remote-review>[]</remote-review>
```

• ○ Verify (not started)
• ○ Dedupe (not started)

So Find ran for ~10 seconds, returned no findings, and the run halted as failed. The quota counter still went from "Free ultrareview 1 of 3" to "Free ultrareview 2 of 3" on the next invocation.

Likely root cause (best guess)

The diff was 692 files / 297,951 insertions because the main base branch was an empty initial commit (the repo had no prior history — first push). Reviewing "everything as additions" appears to be outside Find's working envelope. A subsequent slot used on a smaller scope (27 files, 6.5K lines) succeeded and returned 8 high-quality findings.

Suggested behavior

Either:

1. Don't decrement quota on a Find failure (no findings produced ≠ a successful review).
2. Or detect "diff too large / unreviewable scope" earlier and refuse to launch (clearer feedback than a silent empty-array failure).

Ask

Please credit back the 1 free slot from this failed run. Happy to provide more detail if useful.

Thanks!