Identity
Production self-serve uses verified Google identity, keyed on the Google subject and never on email.
Security
Controls enforced in the product today.
No roadmap promises on this page — only what the running system does now. If a control isn't live, it isn't here.
Tenancy
Postgres row-level security is forced on tenant tables, so a query can only ever see its own workspace.
Readouts
Content-Security-Policy and sandbox headers constrain every deployed readout (a static web page); the default policy is deny.
Data
Every connector's operations are allowlisted and read-only, each with its own read allowlist (Jira shown as the worked example); live reads use the viewer's own account.
Secrets
Every deploy is scanned for secrets and external origins before a version is committed.
Audit
Deploy, sharing, connection, token, edition, billing, and admin actions are all recorded.
Certification
The canonical view is named.
When a steward certifies a readout, the seal records who decided and at what scope. Certification auto-flags the moment the readout stops being healthy — a gold badge means something because it is scarce.
For security and IT teams
The sanctioned place to point your makers.
Your engineers are already shipping AI-built tools. Aptarium is the sanctioned place to point them — every readout in one legible place: who published it, on which data, to which audience. Every edition carries a byline — a stable, dated copy the whole team opens, naming whose access it went out on, and as of when — live reads carry the viewer's own access, there is no standing shared credential anywhere, and nothing is public by default. See the threat model, the data flow, and what Aptarium deliberately cannot do.
Responsible disclosure
Report a security concern to support@aptarium.app. Please do not include credentials or customer data in your report.