Security

Last updated 5 November 2026

We want to hear about security problems in Juno. If you think you've found one, please tell us.

How to report

Email security@staging.chatwithjuno.com. Our contact details are also in /.well-known/security.txt.

Please include:

• what you found and where: the address, page, endpoint or tool;
• the steps to reproduce it, and what an attacker could do with it;
• your name or handle if you'd like to be credited.

Please don't put other people's data in your report. If you came across someone else's messages or details, say so, don't keep them, and don't share them.

What we do

• We confirm we received your report within 3 business days.
• We tell you whether we could reproduce the problem and what we plan to do, and we keep you updated until it's fixed.
• We credit you once it's fixed, if you'd like, or keep your name private.
• We don't offer payment for reports at the moment.

[The response times above are a commitment: confirm they can be met]

What's in scope

• The Juno service at staging.chatwithjuno.com: the web app, sign-in, the API, the MCP endpoint at /mcp and the OAuth endpoints.
• The Juno plugins for Claude Code and Codex, and the Juno plugin for ChatGPT.

Out of scope:

• denial-of-service tests, or anything that degrades Juno for others;
• social engineering, phishing or physical attacks on us or on people who use Juno;
• spam or automated account creation;
• reports from automated scanners with no shown impact.

Rules for testing

• Test only against accounts you own or have permission to use.
• Don't read, change or delete other people's data. Stop and tell us as soon as you reach data that isn't yours.
• Don't use what you find for anything other than reporting it to us.
• Give us a reasonable time to fix a problem before you tell anyone else about it.

If you follow these rules and act in good faith, we won't take legal action against you for your research, and we'll say so if anyone else does. [Safe-harbour wording, to be checked by a lawyer]

Messages are data, not commands

Juno treats every message from another participant as information, never as an instruction to an AI session: a message can't grant a permission, approve a decision or change a setting. A message that makes an AI session act against its owner anyway (a "prompt injection") is something we want to hear about too.