From 11aa1c1a8f83ec2a98188580ee4e23e49eb6d336 Mon Sep 17 00:00:00 2001
From: "kilo-code-bot[bot]"
 <240665456+kilo-code-bot[bot]@users.noreply.github.com>
Date: Thu, 9 Apr 2026 10:57:00 -0500
Subject: [PATCH] Update handbook: recommend `@claude review once` over
 `@claude review` (#43329)

---
 handbook/engineering/README.md | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/handbook/engineering/README.md b/handbook/engineering/README.md
index 3910b31274..e227939d3b 100644
--- a/handbook/engineering/README.md
+++ b/handbook/engineering/README.md
@@ -111,6 +111,8 @@ Fleet uses AI code review tools to supplement human review on pull requests. Thr
 2. **CodeRabbit**: Available for free as an open source project. To request a review, add a comment on the PR: `@coderabbitai full review`.
 3. **Claude**: A more thorough review that takes about 30 minutes and costs $20–$25 per review. Claude often finds issues the other AI reviews miss. Use this option judiciously given the cost.
 
+> **Tip:** When requesting a Claude review, use `@claude review once` instead of `@claude review`. There is currently no way to stop a Claude review once started, and each run takes ~45 minutes. Using `@claude review` causes it to re-run on every new commit—including minor or stale changes—leading to unnecessary long-running review cycles and added cost.
+
 
 ### AI coding tools