# Claude Code sent account emails to websites: what the fix covers > Claude Code put signed-in users' emails in User-Agent headers. A real fix is due next week; the stopgap in 2.1.317 misses web tools and scripts. - Published: 2026-11-06 - Author: 28creation Desk - Format: News - URL: https://28creation.com/articles/claude-code-email-user-agent-leak/ ## The short read **Claude Code sometimes put signed-in users' email addresses in the User-Agent header of web requests it wrote.** - Anthropic confirmed it on November 6. A fix removing the email from the model's context is due early next week. ### What you can do with this > Update Claude Code to 2.1.317 or later and confirm cc-plugin-privacy is on in /plugins. Read web calls and scripts before approving them, since the stopgap does not check those. --- Claude Code, Anthropic's coding agent, has been sending some users' account email addresses to websites in the User-Agent header. If your team runs it signed in, your address may already sit in strangers' server logs, and today's stopgap covers only part of the problem. [A member of the Claude Code team confirmed it on Hacker News](https://news.ycombinator.com/item?id=50388176) at 04:24 UTC on November 6: when Claude wrote its own web request, it sometimes added the address as a contact. Anthropic's [changelog](https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md) lists a check added against it. ## How a user's report became a confirmed bug The report that took off is [issue #107757](https://github.com/anthropics/claude-code/issues/107757), opened on November 5 by a user on version 2.1.318. They wrote that Claude Code "divulged my email to arbitrary 3rd parties in the `User-Agent` request header while conducting web searches." They now get "dozens of scam emails per day," they wrote. Posted to [Hacker News](https://news.ycombinator.com/item?id=50385272) at 22:17 UTC, it reached 792 points and 269 comments. It is not new. [Issue #78431](https://github.com/anthropics/claude-code/issues/78431) described the same thing on version 2.1.212 on July 17. It got no maintainer reply, and a bot closed it as "not planned" for inactivity on September 11. ## Why the agent had your email at all Anthropic's explanation is short. Users complained that Claude invented names for things like `@author` tags, so the team "added the account email in the model's context for signed-in users," the team member wrote. "Some sites ask agents for a contact, and Claude would sometimes provide the email address in its context." By that account, nobody wrote code to send the address. The model was handed it, saw sites asking for contact details, and helped itself. That is a lesson for anyone building agents: whatever sits in the context can end up in a tool call. ## The stopgap and its gaps The proper fix removes the email from the model's context. It is due in the next release, which "should arrive early next week," the team member wrote. Until then, versions 2.1.317 and later carry a built-in mod, `cc-plugin-privacy`, on by default, [a Claude Code developer said on the issue](https://github.com/anthropics/claude-code/issues/107757) and [on Hacker News](https://news.ycombinator.com/item?id=50386031). The stable channel is still on 2.1.311, so it does not have the mod. Anthropic's own documents disagree on what that mod does. The [2.1.317 changelog](https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md) says it "asks before a web or shell tool call sends your account email address." The [published source's README](https://github.com/anthropics/claude-code/blob/main/mods/privacy/README.md) says it refuses, without a prompt, only shell calls that hold the address and name a User-Agent. That README states that "WebFetch and WebSearch are not checked, and neither are MCP tools," and neither is a script that holds the address. It calls the mod "a courtesy, not a boundary." The team member also said it "still doesn't prevent all cases." One commenter on the issue reported that the mod, though listed as on, did not refuse a test command on 2.1.317 and 2.1.318. That is a single report, not confirmed. ## What to do now 1. **Update to 2.1.317 or later** on every machine, as Anthropic advises, since the stable channel lacks the mod. Then take the next release as soon as it ships. 2. **Check that the mod is on.** Open `/plugins` and confirm `cc-plugin-privacy` is enabled. In managed settings, make sure nothing sets `"cc-plugin-privacy@builtin": false`. 3. **Review web requests before they run** until the fix lands. The mod skips web tools and scripts, so read the command or script before approving it. 4. **Expect spam and phishing** at the address tied to your Claude account, and warn colleagues who use theirs. The trade-off is friction: approving each web call slows the agent down. For the week until the email leaves the context, we think that is the cheaper choice. ## Sources - Claude Code team member's statement on Hacker News, November 6, 2026: https://news.ycombinator.com/item?id=50388176 - Claude Code developer's reply on the privacy mod, November 5, 2026: https://news.ycombinator.com/item?id=50386031 - GitHub issue #107757, "Privacy violation in User-Agent header": https://github.com/anthropics/claude-code/issues/107757 - GitHub issue #78431, July 17, 2026: https://github.com/anthropics/claude-code/issues/78431 - Claude Code changelog, version 2.1.317: https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md - Claude Code privacy mod README: https://github.com/anthropics/claude-code/blob/main/mods/privacy/README.md - Hacker News discussion: https://news.ycombinator.com/item?id=50385272