If you ran Umbrella, you probably used the Policy Tester. Pick an identity, type a destination, and Umbrella told you which policy applied and whether the request would be allowed or blocked. No live user, no waiting on a log. You checked the change before anyone felt it.
Secure Access doesn't have that yet. A rule change goes live the moment you save it. The only test today is to save the rule, then go read Activity Search against a real user's traffic. People have asked Cisco for a tester, and it is coming (Cisco Community). Until it ships, we built one.
The gap is also wider than it was in Umbrella. Secure Access carries traffic over more paths, and the path changes the answer. A successful check from your own laptop says nothing about a branch coming in over a site-to-site tunnel. Same destination, different source, possibly a different rule.
How Secure Access actually decides
There is one rule list: internet rules, private access rules, and the defaults, in order. The first rule whose conditions match is the one that counts. Everything below it, for that stage, is never evaluated.
Inside a single rule there is one more split. The rule action, allow or block, is not the same thing as what a security profile does afterwards. A rule can allow the connection, and a web security profile, a DNS security category, file inspection, DLP or an intrusion signature can still stop it. Those decisions are made on the content, not on the destination you typed.
What the checker shows
The Secure Access Policy Checker is a Chrome extension that sits on the dashboard you are already signed into and evaluates that tenant's rules locally, in your browser.
You pick three things:
- The connection the traffic arrives on
- The source (optional) that connection actually carries
- The destination
It then shows the path, stage by stage, and ends in one of four outcomes:
- Allowed
- Blocked at a named stage, with the rule that did it
- Allowed, but steered out of the web proxy
- Bypassed entirely
If a rule depends on a content category or a threat category and the checker doesn't have the answer, it asks you. It does not guess.
The recording below walks through exactly that on a Cisco demo tenant.
Rules that can never fire
Order is the whole mechanism. At each stage, the first matching rule runs and the walk stops. A later rule never gets a turn.
That is how a block dies without anyone disabling it. Someone adds a block for a specific destination, low in the list, for one group. An earlier allow for the parent domain already covers those same people. The block is in the table. It is enabled. It never runs, because the allow matched first. Looking at the block on its own, you would swear it is protecting that group. It is not.
What it does not do
Be clear about the edges before you rely on it:
- It is read-only. It does not create, edit or delete a rule. It reads the policy your session already has.
- It walks rules, not content. Blocks from security profiles (web security, DNS categories, file inspection, DLP, IPS) depend on the actual content, so the checker doesn't show them and doesn't pretend to.
- Zero Trust Access is out of scope in this version. If the agent-based or browser-based ZTA path is the only route to the destination, you get no verdict.
- It is not a Cisco product, and Cisco does not support it. A dashboard change on Cisco's side can break it, and the fix is ours to ship.
Try it
The code is on GitHub: github.com/technoxi/cisco-secure-access-policy-checker. Install steps are in the README.
When Cisco ships a native tester, use it. Until then, check the rule before you save it, not after a user finds it for you. If it gets a verdict wrong on your tenant, open an issue with the rule order and the path you tested, and we'll fix it.