Connector actions
Not every response is containing an endpoint. When a hunt finds a compromised account, a malicious IP or a host another product manages, Scout can propose an action in that product through its connection — revoke a user's sessions in Okta, deactivate an AWS access key, contain a host in CrowdStrike Falcon — or block the IP on the endpoint.
These are proposals only. Scout drafts the action with its target and reason, and a person runs it with one click. No response policy can make them run on their own.
Available actions
| Product | Action | Target | Reverse |
|---|---|---|---|
| Okta | Revoke Okta sessions — signs the user out everywhere. | Okta user id | None. The user signs in again. |
| Okta | Suspend Okta user | Okta user id | Unsuspend Okta user |
| AWS IAM | Deactivate AWS access key | Access key id and IAM user name | Activate AWS access key |
| CrowdStrike Falcon | Contain host (Falcon) — cuts the host off the network in Falcon. | Falcon device id | Lift containment (Falcon) |
Each of these uses the product's existing connection and its credentials. The connection's credentials need the right permissions in the product, for example to manage users in Okta or access keys in IAM. If they don't, the action fails and says why.
Account and network actions on the endpoint
Two saved Actions do the same job on the endpoint itself, through the endpoint agent. They are proposals too, never covered by a policy:
| Action | What it does | Reverse |
|---|---|---|
| Block IP | Blocks a malicious IP at the endpoint's host firewall. | Unblock IP |
| Disable local account | Locks and expires a local account, for when there's no identity provider connected. | — |
How Scout proposes them
In an autonomous hunt that finds something, Scout first collects evidence and contains the endpoints a policy covers. It then proposes the account and network actions the evidence points to, for example suspending the account that ran the malicious process in Okta and blocking the IP it talked to.
Each proposal shows:
- the product and the action;
- the target, such as the user or the access key;
- under Sends:, the request it will send to the product;
- why Scout proposed it, with a link to the evidence;
- the reverse action, where there is one.
Proposals appear in the hunt, in Scout activity as Proposed, and in the Inbox under Needs you. Who may approve follows your organisation's approval rules.
Run or reject a proposal
- Open the proposal from the hunt, Scout activity or Needs you.
- Check the target and the evidence.
- Click Approve to run it, or Reject to turn it down.
Where the product supports it, the action does a dry run first and checks it worked afterwards, as endpoint actions do. The run, its check and who approved it are recorded on the hunt and in the audit trail.
To reverse an action, run its reverse Action on the same target, for example Unsuspend Okta user after Suspend Okta user.
Why these are never automatic
Suspending the wrong account or cutting off the wrong cloud key can stop a business in a way that isolating a test endpoint doesn't. These actions also reach outside the estate a response policy is scoped to by endpoint tags. So in this version, a person always decides, and Scout's job is to have the proposal ready, with its evidence, by the time you look.
Next steps
- Response policies — what Scout may do on its own, and what it never does
- Connections — the connections these actions use
- Scout activity and the morning summary — proposals waiting for you