sremcp

Legal

Privacy policy

Effective October 18, 2026

In short

  • We keep what the Service needs: your account, your organizations, the servers you connect, and your investigations with each question, answer and tool call.
  • The agent reads from the systems your organization connects and sends what it finds to an AI model (Gemini or Claude) to answer you.
  • Credentials are encrypted and never go to the AI model. We don’t sell your data or train models on it.
  • Ask us for a copy of your data, or to delete it, at privacy@sremcp.com.

1. Who we are and what this covers

sremcp (“we”, “us”) runs the sremcp service: this website, the documentation at docs.sremcp.com, the web console at app.sremcp.com and its API, and the sremcp iPhone app (together, the “Service”). This policy explains what we collect, why, who processes it for us, how long we keep it, and the choices you have.

sremcp can also run on your organization’s own infrastructure. A self-hosted deployment is operated by your organization, which decides how it handles data; this policy doesn’t cover it.

2. What we collect

Your account. Your email address and name, and how you sign in: a password, an emailed code, Google, Apple, or your company’s single sign-on. Sign-in is run by Clerk, our authentication provider; we never see your password.

Organizations and roles. The organizations you start or join, your role in each, and invitations: the invited email address, the role, who sent it, and an optional note.

MCP servers you connect. Each server’s name, address and settings, the tools it offers and which of them are approved, and the credentials you give us for it (API keys, headers, OAuth tokens, client secrets). Credentials are encrypted and never shown again — see Security.

Investigations and reports. We keep each investigation — your questions, the agent’s answers, and the tools it called with what they returned — so you can come back to it on any device and your team can pick it up. Each question and each scheduled report is also recorded as a run: the question or report prompt, the answer, the model that answered, how many tools it called, and how many tokens it used. Report schedules keep their name, timing and prompt, and the webhook address the report is sent to, if you set one.

Runbooks and the team’s notes. What your organization writes for the agent to follow, and who changed it when.

Alerts. When your organization connects an alerting tool (such as Alertmanager, Grafana or PagerDuty), we receive and keep its alerts — their names, severity, labels, summaries and links — and the investigations they start.

Slack. When your organization connects its Slack workspace, we keep the workspace’s id and name and an encrypted access token. We receive only messages that mention sremcp or use its command, and read the asker’s Slack email and timezone to match them to their sremcp account. We don’t read channel history.

Feedback on answers. When you rate an answer (thumbs up or down) and say what was wrong, we keep the rating and your note with that run. Your organization’s Owners, Admins, SREs and Managers see them, to improve how the agent is set up.

Tool calls and the audit log. Every tool the agent calls is recorded in your organization’s audit log with its inputs, along with every change in the organization (servers, approvals, roles, invitations, schedules) and who made it. Tool results go to the AI model to produce the answer, and are kept as part of the investigation.

Data from your systems. The agent reads from the systems your organization connects, so tool results and answers can contain whatever those systems hold — log lines, metrics, issue titles, names in them. If your systems hold personal data, the agent may see it. Your organization decides which systems and tools the agent may use.

Technical data. Our servers log each API request’s method, path, status, timing and a request id, not its contents. Our hosting provider processes IP addresses to deliver the Service. On our websites and the web console we use Vercel Web Analytics, which counts page views without cookies. The web console uses cookies to keep you signed in.

Error reports. When the Service or the iPhone app hits an error or crashes, an error report goes to Sentry, our error tracking provider: what failed and where in our code, the page or screen, the app version, and the browser or device model. Error reports never include your questions, the agent’s answers, tool output, credentials, or cookies, and aren’t linked to your account.

The iPhone app. Your sign-in is kept in the iOS Keychain on your device. If you turn on Sign in with Face ID, your email and password are stored in the Keychain on that device only, readable only after Face ID. Face ID itself is handled by iOS: we never receive biometric data. If you turn on notifications, we keep the phone’s push address (a token from Apple, through Expo) and which notifications you want; notifications say what happened, such as “Investigation finished”, and never carry an answer or data from your systems. Signing out removes the phone. The app doesn’t track you, show ads, or use advertising identifiers.

3. How we use it

  • To run the Service: sign you in, show each organization to its members with the access their role allows, run the agent, deliver reports, and keep the audit log.
  • To keep it secure: check every request, prevent abuse, and investigate problems (the request id on an error matches our logs).
  • To contact you: invitation emails, and notices about the Service or these policies.
  • To improve it: from aggregate, non-identifying usage, such as page views and error rates.

We don’t sell your information, we don’t use it for advertising, and we don’t use your questions, answers, or data from your systems to train AI models.

4. AI models

To answer, the agent sends an AI model your question, the conversation so far, the descriptions of the tools it may use, and those tools’ results. The model is Google’s Gemini or Anthropic’s Claude, called directly or through Vercel’s AI Gateway; each run records which model answered. These providers process the data to produce the answer, under their terms for business API use.

We never send your MCP server credentials to an AI model.

5. Who processes it for us

We use these service providers, who process data on our behalf to run the Service:

  • Clerk — sign-in and accounts
  • Vercel — hosting for our websites, the console and the API; web analytics; the AI Gateway
  • Our database hosting provider — stores the Service’s data
  • Google and Anthropic — AI models, as described above
  • Resend — invitation emails
  • Sentry — error and crash reports, as described above
  • Expo — delivering push notifications to the iPhone app, through Apple
  • Apple and Google — when you choose to sign in with them

Data also goes where your organization sends it: the MCP servers it connects receive the agent’s tool calls; report webhooks it sets up receive report output, signed so they can check it came from us; its Slack workspace receives the answers to questions asked there and the findings of alert investigations; and its PagerDuty account receives findings as notes on incidents, when it sets that up.

We may disclose information when the law requires it, or to protect the safety and security of the Service and its users. If sremcp is part of a merger or acquisition, your information may be transferred, and this policy continues to apply to it.

6. How long we keep it

  • Organization data — servers, investigations, alerts, runbooks, runs, report schedules and the audit log — is kept while the organization exists. Whoever started an investigation can delete it. The audit log is append-only by design. An organization’s owner can ask us to delete its data at privacy@sremcp.com.
  • Your account. Deleting it (in the iPhone app: More → Delete account, or by writing to us) removes your sign-in and marks your sremcp profile deleted. What you did inside an organization — its runs and audit log entries — stays with that organization as its record.
  • Request logs are kept for a limited time for operations and security, then deleted.
  • Backups are deleted as they expire.

7. Your rights

Depending on where you live (for example under the GDPR, the UK GDPR or the CCPA), you can ask to access, correct, delete or export your personal data, and to object to or restrict how we process it. You can also complain to your data protection authority.

Write to privacy@sremcp.com. Where the data belongs to your organization — its connected systems, runs and audit log — we may ask it to handle the request, since it decides what goes there.

8. Security

  • Everything travels over TLS.
  • MCP server credentials are encrypted with AES-256-GCM, with a key per organization. They can be replaced but never read back, and only go to their own server.
  • Every request is checked against the caller’s membership and role, and every change and tool call is in the audit log.
  • The agent calls only tools an SRE or admin approved, from servers the asker’s role may use.
  • The iPhone app can lock behind Face ID, and hides the console from the app switcher.

No system is perfectly secure. If a breach affects your data, we’ll tell you as the law requires. More detail is in the security documentation.

9. International transfers

Our providers process data in the United States and other countries. Where the law requires it, these transfers rely on safeguards such as the European Commission’s Standard Contractual Clauses in our providers’ data processing terms.

10. Children

sremcp is a tool for work and isn’t meant for anyone under 16. We don’t knowingly collect their data; if you believe we have, write to privacy@sremcp.com and we’ll delete it.

11. Changes to this policy

We’ll post changes here with a new effective date. If a change is significant, we’ll tell you by email or in the Service before it takes effect.

12. Contact

Questions or requests about your data: privacy@sremcp.com.