Automatic,
never careless.
FlowMeta acts on real websites, so safety is part of the design. These rules are enforced in code, not left to judgment.
Owners only
Every site must be verified with a DNS TXT record or a meta tag before onboarding starts. You cannot connect a site you don't own.
Write guard
Payments and sign-out are never executed. Every non-GET request is checked before it leaves the browser. Delete requests are held back, except ones that remove test records the same run created. Scheduled re-checks are read-only.
Delete only with consent
Delete tools are off by default. If you opt in, FlowMeta generates them and agents must ask their user before every delete call. Revoke it any time and they disappear at once.
Credential isolation
No token passthrough. The AI agent's token never reaches your site, and your site's credentials never reach the agent.
Encrypted secrets
Site credentials, cookies and tokens are encrypted at rest with AES-256-GCM and redacted from every log, including tool call arguments.
Standard OAuth 2.1
Our own authorization server with dynamic client registration, PKCE and resource indicators, so agents act on behalf of users the standard way.
Tenant isolation
Every query on customer data is scoped to its organization, with row-level security for dashboard access.
Hard quotas
Each plan has fixed limits enforced on every tool call. No surprise bills and no overage charges.
Automations
Reports and alerts run on the same tools, with tighter rules.
Read-only runs
Every run gets a short-lived key that can only call tools that read. Tools that change or delete anything are out of its reach.
Only to you
Reports and alerts go only to the owner who created them: to their e-mail and to the Telegram chat they linked with a one-time code.
Your data stays out of the AI
The model that turns a request into a plan sees tool names and the shape of the data, never your site's values. Scheduled runs use no model.
Found a security issue? Email hello@use2mind.com and we'll respond quickly.
FlowMeta