Skip to main content

← Security & Trust

Programmatic access, governed

API and AI-agent (MCP) integrations authenticate with per-application keys — never shared logins. Each key is scoped to least privilege and a single tenant, rate-limited, and revocable at any time, and every request it makes is recorded with the acting identity.

What this looks like in practice

  • API and AI-agent (MCP) integrations authenticate with per-application keys — never shared logins.
  • Each key is scoped to least privilege and to a single tenant.
  • Keys are rate-limited and revocable at any time.
  • Every request a key makes is recorded with the acting identity.
  • The most sensitive operations are withheld from programmatic grants entirely.

Why it matters

AI access is where most platforms get loose. Here it's the most governed path in the system: Claude, Cursor, or any MCP client gets exactly the access you grant, you can see everything it does, and you can pull the plug instantly.

  • Per-app keys
  • Single-tenant scope
  • Rate limiting
  • Instant revocation
  • Identity on every request

Have a security question?

We're happy to walk through our practices in more detail, share documentation, or hear about a concern. A human reads every note.

Email us at hello@visionlabs.com

This page describes our security posture at a high level for transparency; it is not a contractual commitment.