Anthropic adds continuous scanning and fix tracking to Claude Security beta

Continuous scanning works by creating a project, connecting repositories through the Claude GitHub app or an access token, and running a deep baseline scan. After the baseline, Claude scans code changes regularly, keeping findings up to date. A project can hold up to eight repositories, and findings are ranked by severity, include attack‑path diagrams, and can be dismissed or marked fixed; fixed findings reopen if re‑detected. Threat models are generated from code and editable. New role‑based access controls let admins assign Full, Triage or View roles, and the API lets users list projects, resolve findings, and push events to Slack or Jira. Claude Security is also available as an MCP connector in Claude.ai and Claude Code.

The launch offer gives eligible Claude Enterprise organizations up to $25,000 in promotional credits each month for two consecutive months, ending on December 31, 2026. Pricing is token‑based only, with no additional platform fee, and organizations can set spend limits. Anthropic advises that any suggested patches be reviewed by humans before application.