CVE-2026-54316: Claude Code: Out-of-Band Data Exfiltration via Pre-Approved HuggingFace Domain in WebFetch
Because the hostname huggingface.co was pre-approved as a bare hostname for the WebFetch tool, any path on that domain—including attacker-controlled model repositories—was auto-approved without a permission prompt or being subject to --allowedTools restrictions. An attacker able to inject untrusted content into a Claude Code context could direct it to issue WebFetch requests against attacker-controlled repository files (e.g. /resolve/main/config.json), which HuggingFace counts as downloads server-side, creating a covert out-of-band channel for encoding and exfiltrating data Claude can access such as files, environment variables, or command output. Reliably exploiting this required the ability to add untrusted content into a Claude Code context window. Users on standard Claude Code auto-update have received this fix already; users performing manual updates are advised to update to the latest version.
Thank you to hackerone.com/novee for reporting this issue.
Other sources
Claude Code is an agentic coding tool. From 0.2.54 until 2.1.163, because the hostname huggingface.co was pre-approved as a bare hostname for the WebFetch tool, any path on that domain—including attacker-controlled model repositories—was auto-approved without a permission prompt or being subject to --allowedTools restrictions. An attacker able to inject untrusted content into a Claude Code context could direct it to issue WebFetch requests against attacker-controlled repository files (e.g. /resolve/main/config.json), which HuggingFace counts as downloads server-side, creating a covert out-of-band channel for encoding and exfiltrating data Claude can access such as files, environment variables, or command output. Reliably exploiting this required the ability to add untrusted content into a Claude Code context window. This vulnerability is fixed in 2.1.163.
— MITRE
Affected Software
Remediation
Recommended actions to resolve this vulnerability, in priority order.
- Upgrade
Upgrade
npm/@anthropic-ai/claude-codeto a version that resolves this vulnerability.Fixed in 2.1.163 - Upgrade
Upgrade
Claude Codeto a version that resolves this vulnerability.Fixed in 2.1.163
Event History
Frequently Asked Questions
What is the severity of CVE-2026-54316?
The severity of CVE-2026-54316 is rated at 55.
How do I fix CVE-2026-54316?
To fix CVE-2026-54316, ensure that untrusted model repositories are not auto-approved and implement strict hostname verification.
What software does CVE-2026-54316 affect?
CVE-2026-54316 affects the npm package @anthropic-ai/claude-code.
What type of vulnerability is CVE-2026-54316?
CVE-2026-54316 is categorized as an information leak vulnerability.
When was CVE-2026-54316 published?
CVE-2026-54316 was published on June 17, 2026.