Anthropic turns Claude Security into a continuous scanner: findings tracked until the fix ships

• Who / what: Anthropic released an updated Claude Security, its AI-powered code scanner, in public beta for Claude Enterprise customers.
• When: announced October 21, 2026.
• Scale so far: since the April 2026 public beta, Anthropic says thousands of organizations have used Claude Security to find over one million vulnerabilities (company-reported figures).
• What's new: continuous scanning — a deep baseline scan on connecting a repository, then scans that follow each code change based on its risk and complexity — plus fix tracking that marks findings fixed when the fix lands and reopens them if the issue returns.
• Workflow integrations: a new API (list and read projects and findings, resolve or reopen findings, create and manage projects and repositories, personal access tokens, webhooks), an MCP connector in Claude.ai and Claude Code, and role-based access controls.
• Launch offer: $25,000 in free usage credits each month for two months, for usage through December 31, 2026.
• Under the hood: powered by Anthropic's full model family, including Claude Mythos; no Cyber Verification Program access required.

Anthropic rebuilt its AI code scanner into something that behaves like a standing member of the security team. Claude Security now watches your repositories as the code changes, keeps a running list of what it finds, and tracks each finding until the fix actually ships — not just until the report is generated. The update, released October 21, 2026 in public beta for Claude Enterprise customers, is the biggest change to the product since it entered public beta in April.

The timing is pointed: agents now write, review, and ship code on their own, which means the window between a vulnerability entering a codebase and someone noticing it keeps shrinking. A scanner that runs on a schedule is increasingly a scanner that misses things. Anthropic's answer is continuous coverage plus closure tracking — scanning that stays current with the code and a findings list that drains itself when fixes land.

From one-off scans to always-on coverage

"A scan captures one moment, and code changes every day," Anthropic wrote in the announcement. Teams connect repositories through the Claude GitHub app — including GitHub Enterprise Server — or with an access token from GitLab and other Git providers, choosing the branch and scope for each one. Claude Security runs a deep baseline scan, then scans changes regularly based on their risk and complexity, so the findings list stays close to the current state of the code.

Projects can hold multiple repositories, which lets the scanner catch issues that live between services — Anthropic's example is a backend that trusts input a client was supposed to validate, the kind of bug no single-repo scan would see.

Findings that close themselves

Most fixes happen outside the tool that found the issue, which is why vulnerability backlogs grow even at teams that scan diligently. The update changes that: Claude Security checks open findings against the code that changed and marks them fixed when the fix lands — whether the team fixed it in Claude Code or through a separate process. If the issue shows up again later, the finding reopens automatically.

Every finding also gets a severity rating judged against the project's threat model and an explanation of how an attacker could reach the vulnerable code, including a diagram of the attack path — so teams fix what is actually reachable first instead of working down a raw list.

Your team's knowledge shapes the scan

Every environment has controls a scanner cannot see from the code alone. Each project now gets a knowledge base where teams can describe how their code is deployed, which security policies apply, and anything a new security engineer would need to know. Claude Security also generates a threat model from the code itself; the team reviews and edits it, and subsequent scans use it to decide what is worth flagging — and to skip risks the environment already handles.

Plugged into the workflow

A new API lets teams list and read projects and findings, resolve or reopen them, and create and manage projects and their repositories. Users can create personal access tokens that only see and change what the user can, and webhooks notify internal systems when a new finding lands.

Claude Security is also available as an MCP connector in Claude.ai and Claude Code, so an engineer can pull up a finding, ask Claude about it, start the fix, and mark it resolved without leaving the session. New role-based access controls let admins decide who can view projects, triage findings, and create projects — opening the tool to the teams that own the code without handing everyone the keys.

The fine print: beta credits, and what's measured

The update is in public beta for Claude Enterprise customers; admins enable it in the admin console. Eligible organizations receive $25,000 in free usage credits each month for two months, with the offer applying to usage through December 31, 2026.

Anthropic says the product is powered by its full family of models, including Claude Mythos, with no Cyber Verification Program access required — and that since the April beta, thousands of organizations have used it to find over a million vulnerabilities. That is a company-reported count of discoveries, not a tally of confirmed fixes; adoption and verified remediation are the numbers that will show whether the workflow sticks. Either way, the scanner-as-a-continuous-teammate model is now the baseline competitors will have to answer.