Anthropic Testing Bugcrawl Bug Detection Tool for Claude Code

Anthropic appears to be testing a new tool called Bugcrawl inside Claude Code. Opening the dedicated item in the side navigation brings up a repository selection UI, along with a warning that token consumption is very high. The guidance to start with small repositories suggests that this feature runs a substantial amount of analysis in the background.

The most likely role is for Claude to scan the entire codebase, find general bugs, and suggest fixes. The Security tab in Claude Code for Enterprises already focuses on vulnerability detection, and Bugcrawl looks like the next step that covers the broader area of quality and consistency.

There's also room for expansion.

• It could evolve into end-to-end testing that launches a local app and follows user flows.
• How to inject feature specs or test criteria is still an open question.
• For now, it's not visible in production builds, and no public release timeline has been confirmed.

Given the token consumption warning, the first beneficiaries are likely to be Team and Enterprise plans. Anthropic has rolled out Claude Code Security and Claude Code Review in succession over the past few months, and Bugcrawl looks like part of a flow that fills the gap between code security and PR review with bug detection. On the competitive front, OpenAI Codex, xAI Grok Build, and Google Jules continue to compete in repository-level agents.