CodeRabbit

CodeRabbit is an AI code review platform for pull requests, IDE reviews, CLI reviews, CodeRabbit Plan, and Slack-based engineering workflows. It connects to GitHub, GitLab, Azure DevOps, and Bitbucket surfaces, summarizes pull requests, comments on changed code, supports linters and SAST tools, and adds workflow features such as Jira/Linear integrations, autofix, docstring generation, MCP connections, issue planning, and usage-priced Slack agent work.

The June 26, 2026 recheck found no public price change. Pro remains $24/developer/month on annual billing, Pro+ remains $48/developer/month on annual billing, the usage add-on remains $1 per credit for four reviewed files, and the public pricing page still lists CodeRabbit Agent for Slack at $0.50 per active agent minute. The buyer risk is not the sticker price. It is how many review loops agents, humans, and automation trigger in a week.

The best mental model: CodeRabbit is not an AI IDE. It is a review layer for teams already using pull requests.

CodeRabbit’s Knowledge Base is the feature that keeps it from being just another comment bot. It can learn repository preferences from review feedback, detect team rules from agent instruction files such as .cursorrules, CLAUDE.md, and .github/copilot-instructions.md, use linked issue and PR context, connect MCP servers, automatically link related repositories when configured, and search the web for external documentation. Teams that do not want retained Knowledge Base data can opt out in configuration, but that immediately removes stored learnings while stateless sources such as code guidelines and web search continue to work.

What It Actually Is

CodeRabbit reviews code where teams already review code: pull requests. The paid product layers in linter and SAST support, linked repository analysis, analytics dashboards, docstring generation, autofix, pre-merge checks, MCP connections, and issue-tracker integrations.

Pro+ expands the workflow beyond comments. It adds upstream and downstream actions around the review process, including issue planning, CodeRabbit Plan, unit test generation, simplification, merge-conflict resolution, custom pre-merge checks, and other pre/post-merge actions. CodeRabbit now frames review limits as refillable per-developer allowances with fair-usage timing, not a simple hourly reset. Pro includes 5 PR, 5 IDE, and 5 CLI reviews per developer before waiting for refill or using credits; Pro+ raises those to 10/10/10; Enterprise lists 12/12/12. Paid teams can add usage-based review capacity when they intentionally want uninterrupted CLI and PR review loops.

The Slack agent is a separate surface for investigation, recurring tasks, planning, code generation, opening pull requests, and progress reporting from Slack. The current pricing page lists it at $0.50 per active agent minute, and the Agent docs say it is billed independently from CodeRabbit review subscriptions, so buyers should model it as usage spend rather than a fixed review seat.

That means CodeRabbit works best as a pre-review triage layer. It can summarize what changed, flag suspicious areas, enforce known rules, and make simple improvement suggestions. The human reviewer still owns product behavior, rollout safety, threat modeling, and whether the patch should exist at all.

Review Workflow Fit

When To Pick CodeRabbit

• You review many small and medium PRs. AI summaries and first-pass comments save the most time when review volume is steady.
• You maintain public repos. CodeRabbit documents free reviews for public repositories and an OSS plan with Pro+ features under separate rate limits.
• You want PR review across IDE and CLI contexts. Developers can run review before a pull request exists.
• You want review plus automation. Pro+ moves toward CodeRabbit Plan, issue planning, unit test generation, simplification, custom pre-merge checks, and merge-conflict support.
• Your team works in Slack. CodeRabbit Agent can investigate incidents, plan work, generate code, open PRs, and summarize progress from Slack, but it is billed by active agent minute.
• You need enterprise deployment options. Enterprise includes self-hosting, SSO, RBAC, audit logging, and API access.

When To Pick Something Else

• Coding agent inside an IDE: Cursor, Augment Code, or Windsurf.
• Terminal coding agent: Claude Code or Aider.
• GitHub-native assistant: GitHub Copilot if your team wants code completion, chat, and review in one Microsoft/GitHub contract.
• Security-first scanning: use dedicated SAST, dependency, and secrets-scanning tools alongside CodeRabbit.

Pricing

Pricing via CodeRabbit pricing and CodeRabbit plans docs:

Current feature limits worth budgeting around: Pro lists 5 MCP connections and 1 linked repository, Pro+ lists 15 MCP connections and 10 linked repositories, and the plans docs list 20 linked repositories for Enterprise. Teams with many service repos should check these limits before assuming one code-review subscription covers every cross-repo dependency graph.

June 2026 Product Notes

• CodeRabbit Plan in VS Code: the June 10 changelog says the Plans tab can create agent-ready coding plans from workspace files, specs, PRDs, and screenshots, then hand off selected phases to an AI coding agent.
• CLI v0.6.0: the June 9 changelog added coderabbit review --fast, clearer auth/rate-limit status, deprecated --interactive and --prompt-only in favor of plain review mode and coderabbit review --agent, and improved attribution for custom/self-hosted remotes.
• Automatic repository linking: the June 9 changelog says CodeRabbit can discover related organization repositories automatically when enabled, supplementing manual linked-repo entries.
• GitLab and GHES coverage: June changelog entries added the newer CodeRabbit Review experience for GitLab.com/self-managed GitLab and GitHub Enterprise Server surfaces.

Against The Alternatives