Claude Code Patches Exposure of User Emails in HTTP Headers

Summary
The article reports that Claude Code sent authenticated users’ email addresses, and sometimes project-directory names, in HTTP User-Agent headers during web searches and curl requests. Anthropic released fixes in versions 2.1.317 and 2.1.321.
Key points
- Claude Code reportedly included authenticated users’ email addresses in HTTP User-Agent headers during automated web searches and curl commands.
- The article says some headers also contained project-directory names, potentially linking repositories with employee email accounts.
- It attributes the behavior to account emails being added to model prompt context and then reused in contact fields for remote websites.
- The article reports that several affected developers received waves of fraudulent phishing emails.
- Anthropic reportedly added a default-enabled privacy plugin in version 2.1.317 and blocking rules for certain shell commands in version 2.1.321.
- Anthropic pledged to remove account email addresses from future model context; developers were advised to upgrade and audit external network scripts.
Article Details
- Vulnerability Types
- Exposure of account email addresses in HTTP User-Agent headers and outgoing network requests
- Exploitation Status
- reported
- Exploit Availability
- unknown
- Patch Status
- available
- Workarounds
- Upgrade local installations.
- Audit external network scripts for sensitive data exposure.
