Doogree
Response · Remediation

Fix With Consent: Remediating Without Crossing the Line

07/06/2026 · ~5 min read
RemediationNIST RespondTOFUSigned Control Channel

Detection is only half the job. The other half is fixing — but blind automatic remediation is exactly as dangerous as the problem itself. Doogree's approach is simple: we don't fix behind your back. Every fixable issue surfaces as a suggestion, someone approves it, and only then does a known, constrained, reversible action get on its way.

The anatomy of a fix suggestion

Every fixable issue doesn't run on its own — it is proposed. The operator sees three buttons: Approve / Fix, Not now, Dismiss. Tapping Approve sends a modeled, allow-listed action — not an arbitrary shell command — and then the system re-verifies the state and runs the loop again until the issue is resolved or dismissed. There's no "fix everything with one click" running in the dark; there's a short chain of suggestion → approval → action → verification.

A suggestion, not an execution. The default is always that the human (or later — a defined policy) is in the loop. A meaningful action never happens without explicit consent, and each one has a full audit trail — who approved it, what was sent, and what the result was.

Why there's no free shell — the trust model

The secret is that the actions are not "run this command." Actions against a server are sent over a signed, TOFU-pinned control channel (Trust On First Use): the canonical form of the command is signed, so a replay or tamper is detected immediately. There's no free shell — only actions from an allow-list, called by their explicit names. The practical meaning: even if a cloud or a database is breached, an attacker cannot forge a command and turn it into RCE on the fleet. The line between "detect" and "run code on a device" is guarded zealously.

Least-privilege in action. Every action knows exactly what it's allowed to do, and nothing more. There's no "key to every door" — there are specific, allow-listed keys, each of which opens one door only and leaves a trail.

Critical incident → containment button

Fix-with-consent connects directly to the incident engine. When the SIEM-lite engine folds multiple alerts into a correlated incident at high/critical severity, it automatically offers a containment button. Approving the button acts over the same signed control channel — the same trust model, the same allow-list, the same audit. So the attack chain isn't just detected and tagged — it gets a reversible response action one tap away.

Why this is critical for a small business

A small business usually doesn't have a full-time security engineer. Fix-with-consent gives a non-expert business owner a safe "fix this" button — fully documented — instead of needing an expensive SOC. And when it's something meaningful, the human stays in the loop; later you can gate it behind a policy with 2FA approval (push-approval) so that a risky action goes out only after verification.

The theoretical framework: this is the Respond stage of NIST — not just detecting, but responding in a controlled way. Everything is built on least-privilege and on supply-chain safety: there's no channel through which a breached cloud could throw arbitrary commands into the devices. Remediation — yes. Crossing the line — never.

← Back to the blog · How it works →