It is reachable by URL only —
/cache/explorer — and is not linked from
the application menu. It is an operational tool, not part of the end-user
navigation.Who can use it
Only users that are admin or hold a role with access to the FeatureDesigner feature. Both the view and the underlying endpoints enforce this: a user without those permissions is redirected to the home page and the API rejects the request.What you see
The explorer lists one row per cached key:
Above the table, two counters show how many keys are cached and how much memory they use against the limit of the application plan.
Some keys carry the identity of a user in their own name — the email of the session they belong to, a user id, or the text somebody typed into the AI agent. Those keys are shown redacted: the prefix stays readable and the identifying part becomes a short digest, e.g.
USER_VALIDATION-<redacted:9f2c1a04b7de>. You can still search and delete them by that redacted name.
What you can do
- Search — filter the list by any part of the key. The search is a plain, case-insensitive substring match.
- Delete one entry — removes a single key. The next request for that data reads from the source again.
- Clear the whole cache — flushes the cached data of the instance. Safe but not free: the following requests re-read everything from MongoDB / SQL Server, so expect a temporary slowdown.
Clearing does not drop the authentication entries. Rebuilding those means
one Auth0 profile call per active user, so they are deliberately kept: a
repeated flush would otherwise exhaust the Auth0 quota and lock everyone out
of the application. Signed-in users stay signed in.
The cache lives in the memory of each application instance. Clearing it
affects the instance serving your request; if your application runs on several
instances, each one keeps its own cache.