Known CVEnpm/@anthropic-ai/claude-code
CVE-2026-54316: Claude Code leaks data through a pre-approved domain in WebFetch
Medium
Prompt-injected content could make Claude Code send data out, a character at a time, through a domain WebFetch trusted without asking (CVSS 6.0). Shown at Black Hat USA 2026 leaking a CI workflow's API key. Fixed in 2.1.163.
Affected and fixed versions
Affected: >= 0.2.54, < 2.1.163. Fixed in 2.1.163.
Sources
See every known CVE, and the ones in the news, in the Learning Center