Trust model
This page is for whoever has to approve installing Seylo in a cluster they are responsible for. It states what we can reach and what we cannot, and it is deliberately specific about the worst case.
What Seylo holds
Section titled “What Seylo holds”| Cloud keys | None. Seylo stores no cloud key, secret or certificate for any cloud. There is nothing to leak |
| Cloud credentials for agent-managed clusters | None. Connecting a cluster through the agent needs no cloud access at all |
| Access to a cloud account | Only if you connect one. A cloud account is optional, read-only and keyless: you grant a read-only role that trusts Seylo’s identity, and you revoke it by deleting that role |
| A network path into your cluster | None. The connection is outbound-only, from you to us |
| Your Secret values | None. No surface returns one — see how secrets are handled |
| Shell access to your containers | None. No interactive shell and no pods/attach. pods/exec exists only if you set the chart value sre.exec.enabled=true and a Seylo admin opts the cluster in, and then only for a fixed list of read-only commands. See commands in containers |
| Your workload inventory, manifests and history | Yes — that is the product |
How the connection works
Section titled “How the connection works”The agent runs inside your cluster and dials out. Your cluster accepts nothing inbound: no Service, no Ingress, no NodePort, no container port.
The agent authenticates to Kubernetes with its own ServiceAccount, using the in-cluster configuration. Seylo never holds an AKS, EKS, GKE or OKE credential for that cluster.
This is why no firewall change is needed, and it is why platform teams allow the agent at all.
The worst case, stated plainly
Section titled “The worst case, stated plainly”If Seylo’s central server were fully compromised, an attacker would get: the
ability to issue commands within the agent’s RBAC, on enrolled clusters. With
the deploy posture that means patching images, applying the managed manifest
kinds, editing ConfigMaps, restarting Deployments and creating namespaces. With
no posture set, the chart’s defaults also turn on tier 1 and the pod tier,
which include delete on the kinds an application team edits and on pods, and
read, write and delete on Secrets. See
Secrets in the default grant.
They would not get:
- any cloud key — we store none;
- any write to a cloud account you connected, as long as the role you granted stays read-only — Seylo refuses to connect a role that can change anything or read your data, and checks again each time someone presses Test;
- any Secret value — the refusal is enforced in the agent before serialisation and again on the server, so a compromised server cannot ask for one;
- delete, anywhere — with the
deployorread-onlyposture; - attach, or an interactive shell — no code path and no RBAC rule exists;
- exec, on a cluster you have not opted in — the RBAC rule is absent, and the agent refuses on its own even when asked;
- access to non-managed namespaces, when you install namespace-scoped;
- access to any cluster whose agent you have revoked or uninstalled.
Defence in depth on top of RBAC
Section titled “Defence in depth on top of RBAC”The agent also enforces its own scope. A command naming a namespace outside its watched list is refused before any API call is made. So even a compromised Seylo server cannot widen the blast radius beyond what you configured, whatever it sends.
Cloud accounts are separate, optional and read-only
Section titled “Cloud accounts are separate, optional and read-only”Connecting a cluster never needs a cloud account. If an admin connects one
to see what runs in an Azure subscription, an AWS account or a Google Cloud
project, three things hold:
- No key is stored. You grant a role in your cloud that trusts Seylo’s identity. Seylo signs in as itself and receives short-lived access from your cloud each time. There is no key on Seylo’s side to rotate or to leak.
- Read only, and checked. Seylo refuses to connect a role that can write, or that can read the contents of your data, such as storage objects or secrets.
- Metadata only. Seylo never reads a secret value, a key, a token or an app setting. On Azure the grant is the Seylo Reader role, which is Reader minus every read Microsoft documents as returning one; on AWS the stack denies the reads that return instance user data and Lambda environment variables. See what Seylo reads in your cloud.
- You revoke it, in your cloud. Deleting the role assignment, the CloudFormation stack or the service account ends Seylo’s access, whatever Seylo does.
If Seylo’s central server were fully compromised, an attacker could use Seylo’s identity to read what that read-only role allows in each account connected to it — resource names and settings, metrics, logs, the activity log and the bill — and nothing more: no secret value, key, token or app setting, and no write. Not connecting a cloud account removes that, and changes nothing about your clusters. See cloud accounts.
Credentials, and how they are stored
Section titled “Credentials, and how they are stored”Two secrets exist in the agent relationship, and neither is stored in plaintext by Seylo:
| Enrollment token | Per-agent credential | |
|---|---|---|
| Lifetime | ~1 hour, single-use | Long-lived |
| Where the real value lives | Your cluster, in the Secret you installed with | Your cluster, in the agent’s own Secret |
| What Seylo stores | A hash | A hash |
A token cannot be recovered from its hash. The enrollment token is never written to the audit log, and the install page issues it over POST so it never appears in a URL — and therefore never in a reverse proxy’s access log.
Revoking an agent is immediate and is done from the same page that installed it.
Coexisting with Argo, Flux and Helm
Section titled “Coexisting with Argo, Flux and Helm”A change to an object that Helm, Argo CD, Flux or an operator owns is refused with HTTP 409. The refusal names the controller, quotes the evidence — which annotation, label, ownerReference or field manager proved it — and states the consequence.
Proceeding requires an explicit acknowledgement, which:
- records who acknowledged and when, with the full ownership report attached to the approval, so the approver sees the owner before approving;
- writes a dedicated audit row — also written when a change is blocked at approval time, because an owner appeared after submission;
- is re-checked inside the mutating engine, so an approved-hours-later request or an unattended scheduled deploy cannot slip past a new owner.
Enforcement is server-side at three points — submit, approval, and the engine. The interface only explains the refusal.
An ownership check that cannot be performed — an older agent, or missing RBAC — does not block, but is recorded as a warning. It is never recorded as “nothing owns this”.
What we will not add later
Section titled “What we will not add later”These are design positions, not unimplemented features:
- A shell in your containers. A shell cannot be dry-run, diffed, version-recorded, verified or reviewed, and it reads Secret values trivially. The opt-in commands in containers are the alternative: a fixed list of read-only commands, built by the agent from a table, never through a shell, with output redacted in your cluster.
- Returning Secret values. Including to an administrator, including in an export.
- Inventing a number. An unknown cost or metric is reported as unknown, with the reason. It is never estimated from a similar workload.
- Acting outside an approved envelope. Autonomy exists only inside a signed, bounded, expiring and rate-limited grant that you issue.
What to check before approving an install
Section titled “What to check before approving an install”- Read the RBAC that will be created — the install page renders it before you
apply anything, and
helm templaterenders it without installing. - Verify the image and chart signatures.
- Decide the posture and scope.
deployplus namespace scoping is the tightest useful combination. - Confirm the egress rule you are adding: one host, port 443, outbound.