Accounts and Access

Your account is yours. Every request that reaches SenseLab is checked — who is asking, whether they are allowed, and what it counts against — and every route in, whether an MCP client, an SDK or a direct API call, goes through the same checks. There is no side door that skips them.

​

Your data is isolated from every other account

Isolation between accounts is enforced at the storage layer, underneath the application. One account cannot see another’s memory even if there is a bug in application code, because the separation does not depend on the application remembering to filter — it holds whether or not the code above it behaves.

​

What each member can do

Everyone you invite holds one role, and the role decides what they can reach: Rooms have their own lighter roles for collaborators and viewers, covered in Room Members and Permissions.

​

Every agent gets its own key

Each agent or tool connects with its own API key, and a key can be limited to part of your memory rather than all of it — one agent scoped to your checkout service, another allowed to read shared patterns and nothing else. That limit is enforced on every request, not left to the agent to respect. An agent cannot reach outside its scope even by searching for something: memory it was not given access to simply is not there as far as it is concerned. This is what permissioned inference means, and it is why giving an agent a key is safer than giving it your whole store and hoping. Keys are created under Settings → API Keys, where you also set the paths and whether the key can read, write, or both. A key is shown once when you create it and only its prefix is kept afterwards, so a lost key is revoked and reissued rather than recovered.

​

What happens when you hit a limit

Requests are limited per key over a rolling one-minute window. Past that, further requests from the same key are refused until the window moves on. The SDKs and the MCP server handle this for you by backing off and retrying, so in normal use a busy agent slows down rather than fails. Admin keys are not limited. If your agents are hitting the limit constantly, that is usually a sign to look at how often they are re-reading the same memories rather than a reason to ask for a higher ceiling.

​

Sensitive actions are recorded

Memory writes, API key creation and revocation, and changes to your account are written to an append-only audit log — who did it, what they did, what they touched, and when. Nothing removes an entry once it is there. Browse it under Settings → Audit.

​

Seats, operations and quotas

Every plan includes a monthly allowance of operations plus limits on seats, API keys, open rooms, and room guests. Usage is metered as it happens and visible under Settings → Usage, so you can see where you stand before you run out rather than after. Billing & Metering covers what counts as an operation and what each plan includes.