Cursor Launches Bugbot: AI Code Reviewer Now Publicly Available

Key Takeaways:

• Bugbot is now publicly available after a beta in which it reviewed more than 1 million pull requests and flagged 1.5 million issues.
  

• The AI tool integrates with GitHub and Cursor IDE to detect bugs, logic errors, edge cases, and security vulnerabilities—especially in AI-generated code.
  

• Bugbot can analyze error messages, suggest code changes, and even run unit tests, reducing manual review workloads across teams.
  

• Pricing starts at $40 per month, with Pro and Teams tiers that include advanced customization and Slack-based issue reporting.
  

• Anysphere, Cursor’s parent company, has raised $900M and counts OpenAI, Discord, and Shopify among its customers.

Bugbot Launches as Developers Rely More on AI-Created Code

Cursor, the AI-enhanced coding environment from Anysphere, has officially launched Bugbot, its automated code review agent. Designed to detect bugs that escape traditional linters—tools that catch syntax and formatting issues—Bugbot focuses on deeper logic errors, edge cases, and security flaws. Bugbot is built for a world where developers—and AI agents—are writing and shipping software at increasing speed.

“Our core product is giving you software engineering super powers, but software engineering goes beyond just writing code in your editor,” Jon Kaplan, an engineer at Anysphere, told WIRED. “Bugbot is one of the ways we’re now stepping out of the editor.”

The tool runs automatically on pull requests (PRs), integrating directly into GitHub and the Cursor IDE, where it surfaces potential issues and offers instant fixes. Bugbot is now generally available following a beta that included thousands of engineering teams and over 1 million PRs reviewed.

“The heads of AI at our larger customers are looking for the next step with Cursor,” said Anysphere engineer Rohan Varma. “Now that [teams are] moving quicker, it’s, ‘How do we make sure we’re not introducing new problems, we’re not breaking things?’”

Designed for AI-Aided Development Workflows

Bugbot is optimized to review code written by both humans and AI agents—a growing need in modern software development. Cursor estimates that 30% to 40% of code on professional teams is now AI-generated, matching similar figures from companies like Google.

Bugbot works by gathering context from PRs, including the intent behind changes, and then analyzing the code with a mix of proprietary techniques and models from leading AI labs. This allows it to surface not just syntax errors, but logic bugs, security flaws, and edge cases that often slip through manual reviews.

The tool’s deeper capabilities include:

• Analyzing error messages to identify root causes

• Step-by-step problem solving based on PR content

• Suggesting specific code changes to fix issues

• Running unit tests as part of the debugging workflow

These features are aimed at developers who rely heavily on AI-assisted coding—or what Cursor calls vibe coding—a fast-paced style where developers generate, edit, and test code in a single continuous loop.

Results from Beta Testing

During the beta period, Bugbot reviewed 1 million PRs and flagged more than 1.5 million issues, according to Anysphere. Internal data showed that over 50% of those flagged bugs were fixed before merging.

In one incident during internal testing, Bugbot predicted its own failure. When Anysphere engineers noticed that the tool had gone silent for several hours, they began investigating a possible outage. The team eventually traced the disruption back to a specific pull request. Reviewing the logs, they discovered that Bugbot had already commented on that PR, warning that the proposed change would break the Bugbot service. Despite the alert, the change had been approved and merged—leading directly to the outage. The episode became a validation moment for the Anysphere team: Bugbot had correctly diagnosed the bug that would take it down—before any human realized it.

Bugbot has since become central to Cursor’s own development workflow.

Bugbot Pricing and Feature Plans

Bugbot is offered in two tiers, both of which include a 14-day free trial:

Pro – $40/month
• Unlimited reviews on up to 200 PRs/month
• Unlimited access to Cursor Ask
• Integration with Cursor to fix bugs
• Access to Bugbot Rules

Teams – $40/user/month
• Unlimited code reviews across all PRs
• Unlimited access to Cursor Ask
• Pooled usage across your team
• Advanced rules and settings

Bugbot Rules allows engineering teams to define custom coding standards and enforce project-specific guidelines. Once an issue is flagged, developers can resolve it directly in the Cursor IDE or launch a background agent to handle the fix asynchronously.

Slack Integration for Issue Reporting

Bugbot is also available as a Slack bot, allowing users to generate GitHub issues from within a chat interface. This feature is especially useful for non-technical team members or developers working from mobile devices.

Some supported natural language commands include:

• authorize – Authenticate with GitHub using a personal access token

• report – Guide the user through reporting an issue

• new issue / feature – Open a GitHub issue or request

• show log – View the last 10 entries in Bugbot’s log

Future updates may include natural language processing (NLP) to improve issue quality, identify duplicate reports, or assist confused users by offering contextual help.

Trusted by Developers at Leading Companies

Several large-scale engineering teams have already integrated Bugbot into their review pipelines—and many say it’s reduced manual workload without compromising quality.

Ankur Bhatt, Head of AI Engineering, Rippling:
“One of our staff engineers told me: 40% of my time goes into code reviews. Bugbot helps give that time back—and lets our best engineers stay focused on high-leverage work.”

David Cramer, Co-Founder & CPO, Sentry:
“I've tried many AI review tools. Bugbot produced less noise, caught real bugs, and just slotted perfectly into our flow.”

Kodie Goodwin, Discord:
“We've had PRs approved by humans, and then Bugbot comes in and finds real bugs afterward. That builds a lot of trust.”

Vijay Iyengar, Engineering Leader, Sierra:
“Bugbot blew us away with the nuance of the bugs it was catching. The generator-verifier gap is real, and Bugbot is incredibly strong at reviewing AI-generated code.”

Cursor's customers include OpenAI, Shopify, Instacart, Midjourney, Discord, and Rippling, among thousands of others. Even Alphabet CEO Sundar Pichai has publicly acknowledged using Cursor for vibe coding.

Competitive Landscape for AI Code Review Tools

Cursor and Bugbot enter a crowded market of AI-assisted development platforms. GitHub Copilot, built with OpenAI, is widely used as a “pair programmer” that auto-completes code and offers debugging assistance. Replit, Poolside, and Windsurf offer full-stack code generation environments. Cline is a growing open-source alternative.

Cursor is built on Microsoft’s Visual Studio Code and taps into frontier models from Google Gemini, Anthropic Claude, and DeepSeek. Some developers now run Claude Code alongside Cursor, using its debugging features to complement code generation.

Still, concerns persist about how reliable AI-generated code really is. In one recent case, Replit made unauthorized changes during a code freeze, accidentally deleting a user’s entire database. The founder later admitted the incident was “unacceptable.” While extreme, it illustrates why tools like Bugbot—focused on AI code quality—are gaining traction.

Q&A: Bugbot and AI Code Review

Q: What is Bugbot?
A: Bugbot is an AI-powered code review tool integrated with GitHub and Cursor. It catches bugs, edge cases, and security issues before code is merged.

Q: What can Bugbot do?
A: Bugbot analyzes error messages, identifies bugs, solves problems step-by-step, suggests code fixes, and can run unit tests as part of its review process.

Q: How does Bugbot integrate with my workflow?
A: It runs automatically on PRs, comments on issues in GitHub, and enables one-click fixes via the Cursor IDE or a background agent.

Q: Can I customize how Bugbot reviews code?
A: Yes. Bugbot Rules lets teams set their own coding standards, best practices, and project-specific rules.

Q: How accurate is it?
A: In beta, over 50% of flagged bugs were fixed before PRs were merged. It reviewed 1 million PRs and flagged 1.5 million issues.

Q: What does Bugbot cost?
A: Pro plans start at $40/month, and Teams plans cost $40/user/month, with a 14-day free trial included.

Q: Can I use Bugbot on Slack?
A: Yes. Bugbot supports Slack commands for reporting bugs, creating GitHub issues, and reviewing activity logs.

What This Means

The public release of Bugbot signals a shift in how modern development teams are managing code quality in an era of AI-assisted software engineering. As tools like Cursor make it easier—and faster—to generate large volumes of code, the need for automated systems that can review, interpret, and debug that code becomes more urgent.

Bugbot is built to address that challenge. Unlike traditional linters or formatters, it reviews pull requests for deeper issues—logic bugs, security vulnerabilities, and context-specific edge cases—and offers actionable fixes inside the developer’s environment. Its integration with GitHub, Cursor, and Slack also reflects a broader industry trend: quality assurance is shifting closer to the point of code creation.

The tool is also tightly linked to Cursor’s concept of “vibe coding”—a workflow in which developers move fluidly between writing, generating, testing, and debugging with help from AI. Cursor’s adoption of vibe coding as a cultural shorthand points to a future where AI isn’t just a co-pilot, but a foundational part of how software is built.

As AI-generated code becomes standard practice across teams, tools like Bugbot will be essential—not only for catching bugs, but for making AI-powered development sustainable, secure, and scalable.

Editor’s Note: This article was created by Alicia Shapiro, CMO of AiNews.com, with writing, image, and idea-generation support from ChatGPT, an AI assistant. However, the final perspective and editorial choices are solely Alicia Shapiro’s. Special thanks to ChatGPT for assistance with research and editorial support in crafting this article.