Security
Security you can inspect, not just trust.
Isolation enforced at the database, tokens that are useless if leaked, and a stance on agent inputs most tools haven't thought about yet.
Tenant isolation in the database
Postgres row-level security with fail-closed policies on every tenant table. No session context means zero rows — isolation doesn't depend on the app remembering a WHERE clause.
Agentjacking defense
Every event and log field returned to an agent is fenced with untrusted-data delimiters and a suspicious flag — a poisoned stack trace can't smuggle instructions into your agent.
Tokens useless if leaked
Agent and upload tokens are opaque random values, stored only as SHA-256 hashes, scoped to a role, and revocable in a click. A leaked token buys an attacker one narrow, auditable door.
Single-use identity links
Email verification and password resets use single-use, hashed, short-lived tokens. A completed reset kills every existing session and signs you in fresh.
Documented self-host check-ins
Product data stays in your network. A periodic license and update check sends only the limited instance metadata listed in the public telemetry disclosure.
Responsible disclosure
Found something? Email security@useokapi.app. We read every report, respond fast, and credit researchers who help us keep the whole thing tight.
Start catching errors before your users do.
One binary and Postgres, on your own infrastructure. A 14-day trial is built in — no signup, no key. MCP on every plan, and a price that never moves.
14-day trial · Sentry-SDK compatible · Runs on your own infrastructure