A brand allowlist is the set of brands one credential is permitted to reach. It is the authorization boundary for brand-keyed tools: ask for a brand that is not on the list and the answer looks exactly like asking for one that does not exist.
Why it matters
Systems that hold many brands fail in a predictable way. Authorization gets checked near the front door, and then some query further in trusts whatever brand name it was handed.
There is a quieter leak too. An error saying not authorized tells the caller the brand exists; an error saying not found does not. Told apart, those two messages let a stranger map a customer list one guess at a time.
An allowlist carried on the credential closes both. The set of reachable brands travels with the request, and every brand-keyed call resolves against it rather than against a name in the request body.
How Jinn treats it
On the Jinn gateway the brand allowlist is the sole authorization boundary for brand-keyed tools. A brand a token cannot access returns a generic brand not found, because whether a brand exists is not something a stranger gets to learn from an error message.
Scope is pinned on the server as well. Self-serve tokens are read-only and cannot be escalated from the client, and what a token can call is visible to the agent itself: the tool listing shows exactly that credential's reach, so nothing is hidden and nothing is implied.
The auth model is public: scope, brand allowlist, and the not-found posture on brands a token cannot reach. See it
Related terms
The boundary travels with the credential.
Every brand-keyed call resolves against the allowlist, and an unreachable brand looks like no brand at all. Read the auth model.