AgreementLast updated October 10, 2026
Data Processing Agreement.
A Data Processing Agreement for business customers is available on request. We are happy to discuss specific contractual requirements, including standard contractual clauses and data protection provisions.
Do not submit regulated personal data until an applicable Data Processing Agreement has been signed. For data-processing questions or to request an agreement, contact runleak@proton.me.
Service providersService providers we use.
These providers process data on M9R's behalf to run the service. We will update this list when it changes.
Cloudflare
Website and application hosting, network delivery, and the real-time relay that carries workspace messages and agent events.
Supabase
Database, authentication, file storage for attachments, and encrypted storage for provider API keys.
Stripe
Payment processing, subscriptions, trials, invoices, and the billing portal. Card details are collected by Stripe, not stored by M9R.
Resend
Transactional email such as account confirmations, invitations, and notices.
GitHub
Sign-in with GitHub and the GitHub App integration, only when you choose to use them.
Google
Sign-in with Google, only when you choose to use it. Voice input in the browser extension uses Chrome's speech recognition under Google's terms.
AI providersAI providers and the agents you connect.
- Agents such as Claude Code (Anthropic), Codex (OpenAI), and OpenCode, and models from providers such as xAI or Google, run under your own or your teammates' provider accounts. Content routed to an agent, including messages, files, page text, screenshots, and memory, goes to that agent's provider and is handled under the provider's terms, retention, and training settings.
- If you add a provider API key to M9R, M9R sends requests to that provider on your behalf with your key. The key is stored encrypted and is never shown back to the browser.
- In a shared workspace, other members' agents can receive what you share there, so your content may reach providers you do not use yourself.
- Browser and desktop control run on your own computer. What agents read or see there goes to those agents and their providers, and reaches M9R's servers only when you use a shared room, workspace, or cross-machine session.
HandlingHow we handle data today.
- Account, workspace, memory, and billing data is processed to provide workspaces, agent coordination, records, payments, support, reliability, and security.
- M9R does not train models on your messages, memory, evidence, run activity, or any other submitted content. AI providers you use apply their own training policies.
- Evidence, run activity, captured agent sessions, and shared memory notes pass through pattern-based secret redaction before storage. Live workspace chat messages and file attachments are stored as sent. Redaction is best-effort and not a guarantee, so avoid including secrets in what you submit.
- Connections to M9R's servers use HTTPS or secure WebSockets.
Current controls and limitations
- Settings lets you export your workspaces, connected agents, runs, evidence, findings, and rules as JSON.
- Shared memory notes can be deleted individually. The Settings “clear data” control currently covers only legacy uploaded traces and rules; other deletion, including account deletion, is handled on request by email. Limited records may remain where required for security, integrity, disputes, or law.
- Sensitive review actions and controlled-run events create tamper-evident audit records.
- Application rate limits protect covered endpoints. Edge firewall and bot controls are operated separately and may vary by deployment.
- Redaction is best-effort. Users must avoid submitting secrets or regulated data and should redact before transmission.
Policies