CodeRabbit 6.7/10 Good · T2
bestaiq
// CODE

CodeRabbit Review (2026): Automated Pull Request Reviews, Weighed Against Mixed Benchmarks and a Past Breach

6.7/10
Good · T2

Capable and widely adopted for automated pull request review, but buyers should weigh its unresolved benchmark disputes and a serious 2025 security incident before committing.

SUB-SCORE SPINE
6.5
CAP
8.0
VAL
8.0
EAS
5.5
PRI
5.5
SUP
8.5
ECO
Independent · ad-free verdicts · we may earn affiliate commissions — this never affects our scores.
FIG · QUICK ANSWER

CodeRabbit is an AI code review tool that comments on pull requests automatically, generating a summary, a walkthrough, and a diagram of what changed. It also reviews code inside the IDE (VS Code and forks like Cursor and Windsurf) and from the command line, before a pull request even exists. The company has raised $88 million across two funding rounds and reports thousands of paying customers, but its own accuracy claims conflict with benchmarks published by competitors, and a serious 2025 security vulnerability is part of its public record.

TABLE · AT A GLANCE
Category AI code review (developer tooling) [liveFacts] ✓Jul'2026
Starting price Free (paid plans from $24/developer/month, billed annually) [liveFacts] ✓Jul'2026
Platforms GitHub, GitLab, Azure DevOps, Bitbucket (beta) [liveFacts] ✓Jul'2026
IDE support VS Code, Cursor, Windsurf [liveFacts] ✓Jul'2026
Founded 2023 [liveFacts] ✓Jul'2026
Funding raised $88 million (Series A + Series B) [liveFacts] ✓Jul'2026
G2 rating 4.8/5 (26 reviews) [liveFacts] ✓Jul'2026
Free tier Yes, rate-limited, no credit card required [liveFacts] ✓Jul'2026

//Strengths & things to watch

STRENGTHS
  • Reviews pull requests automatically and generates a summary, walkthrough, and diagram of the changes, which G2 and AWS Marketplace reviewers cite as a time saver.
  • Works across four major git hosting platforms (GitHub, GitLab, Azure DevOps, and Bitbucket) plus VS Code and its forks Cursor and Windsurf.
  • Free tier covers unlimited public and private repositories with PR summarization and IDE and CLI reviews, with capped rate limits rather than a hard feature wall.
  • In an independent audit of 28 pull requests by an open source maintainer, about 72 percent of its 290 review comments were judged relevant, and about 71 percent of those brought real value.
  • States in its privacy policy that neither CodeRabbit nor its underlying model providers use private repository code to train models; that exclusion does not extend to public open source code, which the company does use for training.
  • Claims SOC 2 Type II and GDPR compliance on its own trust center, and offers a self-hosted option plus a reverse tunnel connector for private git servers.
THINGS TO WATCH
  • A 2025 vulnerability, nicknamed PwnedRabbit by researchers, let an attacker run arbitrary code through an unsandboxed linter and obtain CodeRabbit's GitHub App private key, which could have granted access to roughly 1 million connected repositories. CodeRabbit patched it within days of disclosure, but did not publish a public writeup until about seven months later.
  • Competing vendors publish their own benchmarks, and each ranks itself first. Greptile reports a 44 percent bug catch rate for CodeRabbit versus 82 percent for itself; Entelligence reports an F1 score of 33.0 percent for CodeRabbit versus 47.2 percent for itself; DeepSource reports 36.19 percent F1 for CodeRabbit on a CVE dataset versus its own 84.51 percent. None of these are independently audited.
  • G2 and AWS Marketplace reviewers repeatedly describe support as a chatbot-first flow with delayed responses.
  • Reviewers on G2 and AWS Marketplace also note the tool can generate excessive or irrelevant comments on some pull requests, requiring manual filtering.
  • Chat is only available inside pull requests, not in the IDE, according to CodeRabbit's own documentation.
  • Company headquarters, exact founding date, and employee headcount cannot be confirmed from a single authoritative source; third party listings conflict with each other.

//How it works

  1. 01
    Connect a repository

    Install CodeRabbit as an app on GitHub, GitLab, Azure DevOps, or Bitbucket, or connect the CLI or IDE extension directly.

  2. 02
    Automatic pull request review

    When a PR opens, CodeRabbit generates a summary, a walkthrough, and a diagram of the change, then posts inline comments flagging bugs, style issues, and guideline violations.

  3. 03
    Fix or discuss

    Simple fixes can be applied with one click; harder ones are handed to an AI coding agent. Inside the PR, users can chat with the bot to request unit tests, documentation, or a Jira or Linear issue.

  4. 04
    Review in the IDE or CLI

    The VS Code extension (and Cursor or Windsurf) and a standalone CLI let developers get review comments before a PR is even opened.

  5. 05
    Learn from feedback

    How a team replies to or resolves review threads feeds a 'Learnings' system that CodeRabbit applies to future reviews, according to the company.

FIG · SCORE BREAKDOWN
CapabilityCAP
0.30WEIGHT
6.5
ValueVAL
0.20WEIGHT
8.0
Ease of useEAS
0.15WEIGHT
8.0
PrivacyPRI
0.15WEIGHT
5.5
SupportSUP
0.10WEIGHT
5.5
EcosystemECO
0.10WEIGHT
8.5

EDITORIAL NOTEDownward adjustment reflects the severity of the 2025 "PwnedRabbit" RCE vulnerability (unsandboxed linter execution leaking CodeRabbit's GitHub App private key, risking access to roughly 1 million connected repositories) and the roughly 7-month delay before public disclosure, which the privacy sub-score alone does not fully capture.

SCORING PIPELINE — SHOW THE WORK
SOURCES
64
SUB-SCORES
6 DIMS
WEIGHTED
Σ=1.0
EDITORIAL
+OVERRIDE
VERDICT
6.7/10

//Who it's for

CHOOSE CODERABBIT IF…
  • Teams that want automated first-pass PR review across GitHub, GitLab, Azure DevOps, or Bitbucket without a large budget, given the free tier and the $24 per developer per month entry price.
  • Developers using Cursor, Windsurf, or plain VS Code who want review comments before opening a pull request.
  • Organizations that already keep rule files like CLAUDE.md or AGENTS.md in their repos, since CodeRabbit reads and applies them automatically.
LOOK ELSEWHERE IF…
  • Teams for whom a documented, unsandboxed-execution security incident is disqualifying, regardless of how quickly it was patched.
  • Buyers who need a single, independently audited accuracy benchmark before choosing a tool; none currently exists for CodeRabbit.
  • Teams that rely heavily on responsive human support, given repeated complaints about chatbot-only support.
TABLE · HOW IT COMPARES
ToolScoreTierFromFreeLink
CodeRabbitTHIS TOOL 6.7 Good · T2 $0/mo [liveFacts] Yes
Playwright 8.4 EXCELLENT · T1 $0/mo [liveFacts] Yes Review →
Claude Code 7.9 GREAT · T2 $17/mo [liveFacts] No Review →
GitHub Copilot 7.8 GREAT · T2 $0/mo [liveFacts] Yes Review →

//What users say

AI synthesis of external reviews · not on bestaiq

G2
4.8/526 reviews
[S]
AWS Marketplace
4.7/528 sources
[S]
Capterra
MIXED1 source
[S]
Gartner Peer Insights
4.2/521 sources
[S]
◆ AI SUMMARY

Synthesized from 7 external reviews. Independent signal (Trustpilot / Reddit / verified aggregators) weighted higher than commission-carrying review sites.

MOST PRAISED
  • Ease of setup and PR-review automation cited across G2 and AWS Marketplace reviews
  • PR summaries and auto-generated diagrams called out as the most helpful feature in a Gartner-quoted review
  • Independent maintainer audit found roughly 72% of review comments relevant and 71% of those valuable
MOST CRITICIZED
  • Chatbot-only support with delayed responses, repeated across G2 and AWS Marketplace reviews
  • Excessive or irrelevant comments requiring manual filtering on some pull requests
  • Competitor-run benchmarks (Greptile, Entelligence, DeepSource) all place CodeRabbit's accuracy well below their own product
  • Hacker News criticism of the delayed, perceived-as-templated public response to the 2025 security vulnerability
TAKEAWAYReview-platform ratings are consistently high (4.7-4.8/5) but sample sizes are small (21-28 reviews each) and Capterra shows no data at all

//Frequently asked

Q1

Does CodeRabbit train its AI models on my code?

According to its privacy policy, CodeRabbit and the underlying model providers, stated as OpenAI and Anthropic, do not use private repository code to train models. That exclusion does not apply to public open source code, which CodeRabbit says it does use for training.

Q2

What did the PwnedRabbit vulnerability involve?

Security researchers found that CodeRabbit ran a linter without sandboxing, which let a crafted pull request execute arbitrary code and expose CodeRabbit's GitHub App private key. That key could have been used to access roughly 1 million connected repositories. CodeRabbit disabled the affected linter and rotated credentials within days of being notified in January 2025; the researchers published their findings publicly about seven months later, in August 2025.

Q3

How much does CodeRabbit cost?

There is a free tier with capped usage. Paid plans start at $24 per developer per month billed annually, or $30 per developer per month billed monthly according to CodeRabbit's documentation, with a Pro Plus tier at $48 per developer per month billed annually. Enterprise pricing is not published and requires contacting sales.

Q4

Which git platforms and editors does it support?

CodeRabbit integrates with GitHub, GitLab, Azure DevOps, and Bitbucket Cloud, which launched in beta in February 2025. It also has a VS Code extension that works with Cursor and Windsurf, plus a standalone CLI.

Q5

Is CodeRabbit's accuracy independently verified?

Not fully. The most neutral benchmark found, run by the research lab Martian, reports CodeRabbit had the highest recall among tools tested in its online sample, but the full detailed leaderboard was not accessible for independent confirmation. Benchmarks published by competitors such as Greptile, Entelligence, and DeepSource all rank their own product first and CodeRabbit noticeably lower; those are vendor-run comparisons, not neutral audits.

BOTTOM LINE
CodeRabbit

CodeRabbit is a broadly integrated, reasonably priced automated PR review tool with positive ratings on G2 and AWS Marketplace, but its accuracy claims are contested by competitor benchmarks, and it carries the record of a serious 2025 security vulnerability that took months to disclose publicly.

Good · T2 6.7/10

//Related tools

//Featured in

64 sources· Last verified Jul 2026 ✓ VERIFIED