A bug report filed on November 5 in Anthropic's claude-code GitHub repository says Claude Code included the user's account email address in the User-Agent header of outbound web requests made while searching the web. The issue, #107757, is titled "Bug Report: Privacy violation in User-Agent header" and carries the labels area:security, area:tools and bug.
The reporter, who listed Claude Code version 2.1.318 on macOS, wrote that they "receive dozens of scam emails per day" as a result and asked for the behavior to be "removed from the harness entirely" rather than made a setting. The post drew roughly 700 thumbs-up reactions, and a Hacker News thread about it passed 1,500 points.
A repository contributor replied the same day that a built-in mod called cc-plugin-privacy, introduced in v2.1.317 and later, "deterministically" blocks the behavior. The contributor said the mod is enabled by default and can be turned off via /plugins, and that its source would be published alongside the other built-in mods.
Follow-up comments say the fix is incomplete. One user wrote that when Claude writes its own curl commands or scripts, it can make up User-Agent strings from session details such as project names, and that the mod only matches the account address. That means a script written to a file and run separately can still send it. The same user said a test command containing the account email ran without being refused on version 2.1.319. They asked that model-written requests default to a generic User-Agent such as claude-code/<version>. As of publication the issue remained open.
For teams running coding agents with web access, the report is a reminder that agent-written HTTP requests can carry identifying data from the session. Checking outbound headers, or routing agent traffic through a proxy that strips them, guards against this kind of leak whichever agent is in use.