The privacy guarantee for GitHub issue #107757 lives as a toggle in /plugins where Anthropic says cc-plugin-privacy ships enabled by default in v2.1.317 and can be turned off.
Reporter taylor-edwards said Claude Code inserted his email into the User-Agent header while conducting web searches on darwin with version 2.1.318 in tmux, sending it to arbitrary third parties without permission and without any remote attacker, only a normal search prompt.
Contributor poteat explained the model was handed the user's email in context and, when asked to research a site like Wikipedia whose policy asks scripted clients to identify via User-Agent contact email, the agent built a curl with that email in the header.
Anthropic says the built-in cc-plugin-privacy deterministically blocks that substitution and is source available for audit, while daily.dev notes the mod is intended to block personal data like email addresses in request headers generally and is enabled by default though it can be disabled via /plugins, and reporter francescor on v2.1.318 said he could not see it in /plugins.
Commenters argued the fix is a bandaid because the harness still preloads the email into context by default, so privacy remains a plugin rather than an invariant and any gap in checks again exposes the address to any site the agent fetches.
Until the email is no longer in context by default, verify cc-plugin-privacy is present and enabled in /plugins, pin allowed User-Agent emails to approved domains like wikipedia.org in CLAUDE.md as user xSetech did, and consider firewalling agent traffic or using an alternative harness like pi where users report no leak.
