Control layers
Four boundaries that matter.
A trustworthy agent product cannot rely on a good prompt alone. Identity, separation between accounts, spend, and sending behavior need firm enforcement and human visibility.
01 / IDENTITYOne API key
The API key scopes MCP and API actions to one account. The full key is shown once, and Coldrig keeps only a hashed copy, never a readable one. An app you connect with Sign in with Coldrig holds its own short-lived token with the same power as the key on the MCP endpoint, including spending money, so only the account owner should allow one; you can disconnect it at any time from the dashboard.
02 / SESSIONBrowser session
You sign in to the dashboard with an emailed link or your API key. Your browser then holds an HTTP-only session, so page scripts do not routinely handle your key.
03 / POLICYServer guardrails
Suppression, stop-on-reply, limits, idempotency, and health-based state transitions are service rules, not agent promises.
04 / ACCOUNTSScoped data paths
Every dashboard and tool request is checked against the signed-in account first. Customers do not share a sender reputation pool.
05 / OWNERVisible control
The owner sees health, campaigns, replies, setup, and billing projections. Self-serve billing (Stripe live mode) is active — $49 a month covers the first 3 mailboxes and each extra mailbox is $5 a month — and every real purchase is recorded against a monthly spending limit on the supplier account as it runs, and unusual spend goes to a support check.
06 / DISCLOSURENo hidden claims
A lost API key cannot be read back, and there is no compliance certification claim.