Skip to content
NewA calmer, faster ZeroHour

How Claude Code’s 256KB blind spot put developers’ secrets in the preview pane

CSO Online ()4 min read9 · 6 reads
AI summary · grok-4.7-buildSummaries by Vani AI

Noma says Claude Code's HTML preview could run malicious JavaScript and steal developer secrets; Anthropic patched it.

Noma Security disclosed SneakPeek, a high-severity flaw in Anthropic's Claude Code Desktop preview pane, scored CVSS 8.7 with no CVE published yet. HTML files larger than 256KB were loaded as file:// URLs, skipping the strict content-security policy applied to smaller files, so malicious JavaScript could read local files and exfiltrate AWS credentials, GitHub secrets, and SSH keys. An attacker could plant the file in a public repository and wait for a developer to ask Claude Code to inspect it. Anthropic's first patch was bypassed through a Web Worker; Noma says a deeper fix has fully resolved the issue.

Key facts
  • HTML files over 256KB rendered via file URLs and bypassed the preview CSP.
  • Embedded JavaScript could read local secrets, including cloud, GitHub, and SSH keys.
  • Anthropic assigned CVSS 8.7; a CVE was requested but not yet public.
  • A first patch missed Web Worker reads; a later architectural fix closed it.
Explain this AI news with Vani AIWhat was announced, what is new and why it matters, in a new chat

From the original article

4 min read at csoonline.com

Simply asking Anthropic’s Claude Code to review an HTML file could hand attackers access to a developer’s machine, cloud credentials and software development environment, according to a new Noma Security finding.

A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.