Privacy Policy
Last updated: May 28, 2026
Multi Agent Manager ("MAM", "we", "us") is a desktop application that helps you organize and switch between AI chat tools like ChatGPT, Claude, Gemini, and Copilot. This policy explains what information we collect when you use MAM, how we use it, and the choices you have. It applies to the MAM desktop app, the companion Chrome extension, and this website.
Information we collect
MAM is designed to be local-first: most data about your queue, window positions, and recently used chats lives on your own computer and never leaves it. The information below is the only data that may be sent to our servers, and only when you choose to use the related features.
- Account information. If you sign in to use Team Projects, we receive your email address, display name, and profile photo from Google via OAuth. We store these in our database so your collaborators can see who you are.
- Team Projects data. If you create or join a team project, we store the project name, member list, roles (owner / full / add / view), saved chat metadata (title, source app, save time, optional notes/tags), and any share or collaboration URLs you paste in. The actual conversation content from third-party AI tools is not uploaded to us — only the links and metadata you provide.
- Feedback submissions. If you submit the feedback form on this site, we store the name, email, feedback type, optional app version, message, and your browser's user-agent string. We also store a one-way hash of your IP address (not the raw IP) for spam triage.
- Authentication emails. When you're invited to a project, we send a sign-in link to the email address provided. Email delivery is handled by Resend.
-
Local app data. The MAM desktop app reads
and stores several categories of data locally on your
computer. None of the items in this bullet are transmitted
to our servers — they exist only on your device so MAM can
show you the right thing at the right time.
- Window positions, queue state, tracked Chrome tab references, recently used chats, and your sign-in session token.
- Window titles from other AI applications you have open — for example Claude Code, Codex, Windows Terminal, ChatGPT desktop, Claude desktop. MAM reads these via the Windows accessibility APIs (UI Automation) to detect which chat lives in which window so it can surface the right one when you click on its queue row. Window titles can include whatever you've named your chat (e.g. "Fix authentication bug" or "Plans for Q3 launch"). MAM does not read the body of any conversation, only the title bar text exposed by the operating system.
- Chrome profile metadata. When the MAM Chrome extension is installed, MAM reads each connected Chrome profile's directory name, Google account ID (Gaia ID), and email address through the extension. This is used to route share requests to the correct profile when you have multiple Chrome profiles signed into different Google accounts.
-
Diagnostic logs. If MAM encounters an
unexpected error, a stack trace is written to a local
log file under
%APPDATA%\\multi-agent-manager\\logs\\on Windows so the error can be diagnosed. These logs are never sent anywhere automatically. You may choose to share a log file with us when reporting a bug.
What we do NOT receive: we never see or store the contents of your conversations with AI tools. The Chrome extension observes only window/tab state (title, URL, ready or generating status) — it does not read message bodies. The desktop app's local reads of other applications' window titles, and of Chrome profile metadata, described above, stay on your device and are never transmitted to us or any third party (unless you opt in to crash reporting as described below, in which case scrubbed error data is sent to Sentry).
How we use information
- To let you sign in and identify your team collaborators.
- To deliver the Team Projects feature — showing the projects, members, and saved chats you have access to.
- To send transactional emails (sign-in links, project invites) and to forward feedback form submissions to our support inbox.
- To respond to your questions, bug reports, and feature requests.
- To investigate abuse and keep the service running (e.g., the hashed IP is used to rate-limit spam).
We do not sell your data. We do not use it for advertising. We do not share it with third parties except the service providers listed below, who process it on our behalf.
Third-party service providers
- Supabase — hosts our database, authentication, and the small server functions that power Team Projects and the feedback form. See Supabase's privacy policy.
- Google — handles "Sign in with Google" via OAuth. We receive only the basic profile fields you authorize (email, name, profile photo). See Google's privacy policy.
- Resend — delivers our transactional emails (sign-in links, feedback notifications). See Resend's privacy policy.
- Cloudflare — serves this website and provides DNS for our domain. See Cloudflare's privacy policy.
- Sentry (optional, off by default) — error monitoring. Only used if you explicitly enable it on your machine. See the crash reporting section below for what is sent and how it is scrubbed, and Sentry's privacy policy.
Optional crash reporting (Sentry)
MAM includes an optional integration with
Sentry, a third-party error-monitoring
service. The integration is off by default.
Nothing is transmitted to Sentry unless you explicitly enable
it by placing a Sentry account identifier (a "DSN") into a
configuration file at
%APPDATA%\\multi-agent-manager\\sentry-config.json
on Windows.
If you enable Sentry, the following may be transmitted to Sentry's servers (hosted in the United States) when MAM encounters an unhandled error:
- The error message, error class, and stack trace.
- MAM's version, your operating system, and the Electron runtime version.
- "Breadcrumbs" — a short trail of recent action types inside MAM that preceded the error (e.g. an HTTP request that was in flight).
Before any event leaves your computer, MAM applies a client-side scrubbing layer that:
- Reduces every URL to host-only — paths and query
strings (which can contain chat IDs, share tokens, etc.)
are replaced with
[REDACTED]. - Strips cookies and authentication headers entirely.
- Replaces any string longer than 32 characters that
looks like a token (base64-ish, URL-safe alphabet) with
[REDACTED_TOKEN]. - Truncates exception messages to 500 characters.
- Drops console-log breadcrumbs, which often contain user content, before sending.
You can turn off Sentry at any time by deleting the
configuration file, or by setting
"enabled": false in it. Sentry's own privacy
policy is available at
sentry.io/privacy.
Data retention
We keep Team Projects data for as long as your account exists and you remain a member of the relevant project. Feedback submissions are kept indefinitely so we can refer back to product history; if you'd like a specific submission removed, contact us. Local data on your computer is removed when you uninstall the app or clear MAM's data directory.
Your rights and choices
- Access and export. You can see most of your data directly in the MAM app (projects, members, saved chats). To request a copy of anything else we hold about you, email support@multiagentmanager.app.
- Deletion. Email support@multiagentmanager.app to request deletion of your account and associated data. We will respond within a reasonable time and confirm once the deletion is complete.
- Correction. You can update your display name and profile photo via your Google account, which propagates back to MAM the next time you sign in.
- Sign out. Signing out of MAM clears your local session. You will need to sign in again to access Team Projects, but your project memberships remain on the server until you leave or are removed by an admin.
Security
We use HTTPS for all network communication between the MAM client and our backend. Database access is gated by Row-Level Security policies so users can only see data they are members of. We rely on Supabase, Google, and Resend for transport and storage security. No system is perfectly secure, so we recommend using a strong unique password on the Google account you use for MAM.
Children
MAM is not directed to children under 13, and we do not knowingly collect personal information from children under 13. If you believe a child has provided us with personal information, contact us and we will delete it.
International users
Our servers are hosted in North America. If you use MAM from outside this region, you understand that your data will be transferred to and processed in North America. Where applicable (e.g., GDPR for EEA residents), you have the right to request access, correction, or deletion of your data via the contact address below.
Changes to this policy
We may update this policy as the product evolves. If we make material changes, we will update the "Last updated" date at the top of this page and, for significant changes, notify active users by email or an in-app message.
Contact us
Questions, requests, or concerns about this policy or your data? Email support@multiagentmanager.app and we will get back to you.