Anthropic adds a new feature that automatically detects error codes in Claude Code extremely quickly.

The AI ​​company Anthropic has just added a new feature to its programming assistant, Claude Code , to help developers detect and fix errors faster. The new feature, called Code Review , allows the system to automatically analyze changes in source code, detect potential problems, and suggest fixes before the code is merged.

 

According to Anthropic, when a pull request (PR) is created, Code Review activates a group of AI agents to examine the source code. These agents work in parallel to search for errors, then verify them to eliminate false positives. The detected errors are ranked by severity , helping programmers quickly prioritize fixes.

The analysis results will be displayed directly in the PR as an overview comment , along with detailed comments placed directly on the problematic lines of code.

Anthropic states that its code review system can automatically scale depending on the size and complexity of the pull request. For large changes, the system will use more agents and perform a deeper analysis. Conversely, smaller updates will only require a lighter review process. In internal testing, Code Review typically completes a pull request review in approximately 20 minutes .

 

Anthropic stated that they developed this tool after noticing that the volume of source code generated by each engineer had increased by approximately 200% over the past year . This is largely due to the use of AI-based programming tools.

Currently, the company uses Code Review for almost every internal pull request , and reports a significant increase in the number of truly valuable reviews compared to before.

Following successful internal testing, the Code Review feature is now being rolled out as a research beta to Claude for Teams and Claude Enterprise customers .

However, Anthropic also notes that this tool is significantly more expensive than lighter solutions, such as Claude Code GitHub Action , because the system is charged based on the amount of tokens processed.

According to Anthropic, the average cost per code review using Code Review ranges from approximately $15 to $25 , depending on the size and complexity of the pull request.

To help organizations control costs, the company offers several management tools such as:

1. Monthly spending limits for the entire organization.
2. Set restrictions on a per-repository basis.
3. An analytics dashboard to track the number of PRs reviewed, acceptance rate, and total cost.

The Code Review feature was launched amidst growing interest in Claude Code from businesses. According to Anthropic, the tool's run-rate revenue has surpassed $2.5 billion since its launch, more than doubling from the beginning of 2026.

In addition, the number of business subscription packages has quadrupled since the beginning of the year. Currently, business customers account for more than half of Claude Code's total revenue.