Foundations·27 Oct 2026·6 min read

Keep Your Vault Tooling, Move Your Secrets: Compatibility as a Migration Strategy

Secrets platforms are sticky, but not because of the vault: the lock-in lives in everything pointed at it, from CLI habits to ExternalSecret manifests to ArgoCD plugins. Monopam's answer is a Vault-compatible API over its Secrets Store, so consolidating onto the PAM-governed platform means changing an address, not rewriting a toolchain.

Keep Your Vault Tooling, Move Your Secrets: Compatibility as a Migration Strategy

Ask a platform team why they have never consolidated their secrets, and the answer is rarely the vault itself. It is everything pointed at the vault: years of CLI muscle memory, hundreds of ExternalSecret manifests, the ArgoCD plugin baked into every pipeline, the AppRole wiring in CI. A secrets management platform is sticky not because of the data it holds but because of the integrations that have grown around its API.

Monopam takes that observation seriously. Its Secrets Store exposes a Vault-compatible API, a KV v2 subset that speaks the protocol your tooling already speaks. Consolidating secrets onto the PAM-governed platform then stops being a rewrite project and becomes a re-pointing exercise: the clients stay, the address changes.

The gravity is in the clients

Every team that adopts a secrets store accumulates clients around it. Engineers learn vault kv get. GitOps repositories fill with SecretStore and ExternalSecret resources. Deployment pipelines grow <path:...#KEY> placeholders. None of these artifacts care where the secret physically lives; they care that the endpoint answers in the dialect they were built for. Any migration plan that starts with "first, rewrite all of that" has already lost the room. Compatibility is the migration strategy: keep the dialect, move the data.

Same shape, familiar paths

Secrets in Monopam are organized as projects, environments and folders, and the Vault-compatible API maps a project to what Vault calls a mount. A secret lives at <project>/data/<environment>/<folder>, so the path structure your manifests already use translates directly. Folders and single leaf keys resolve as deeper paths, and reads return version metadata, the way a KV v2 client expects.

One detail teams tend to appreciate: secret references defined in Monopam arrive resolved. If DATABASE_URL is composed from other keys on the platform, the client receives the finished value, not the recipe.

Three clients, zero rewrites

This is not compatibility on paper. Three of the most common clients have been verified against it live, including on a real GKE cluster:

The vault CLI (1.18.3). vault write auth/kubernetes/login role=... jwt=... logs in; vault kv get -mount=<project> <env> reads; kv list enumerates. The binary neither knows nor cares that a PAM platform is answering.

External Secrets Operator, using its native vault provider: the SecretStore reports Valid, the ExternalSecret reports SecretSynced, and Kubernetes Secrets flow as they always did.

argocd-vault-plugin (1.18.1): <path:project/data/env#KEY> placeholders render through avp generate, so GitOps manifests keep their shape.

Machines authenticate the way they already do, too. The API supports Kubernetes login (a service-account token bound to a specific service account), JWT/OIDC, and AppRole with the usual role_id and secret_id pair. The returned token behaves like a client expects: lookup-self, renew-self and revoke-self all work.

What changes underneath

The point of consolidating is not to end up with the same thing under a new logo. Behind the compatible surface sits the PAM platform's machinery: audit on every access, policy, versioning and rollback on every secret.

Access is gated by per-project and per-environment grants on machine identities. A token with no grant on a project cannot even list its environments, or learn that the project exists: the response is the same 403 a nonexistent project returns. A wrong service account is refused by name. A revoked token gets permission denied. And where you need credentials that genuinely disappear, the platform also issues dynamic Kubernetes service-account tokens with real revocation.

The API itself is treated as the privileged surface it is. It ships off by default, is enabled under System > Secrets API, and carries an IP allowlist and login throttling from the first request.

If you would rather go CRD-first

Compatibility is a bridge, not an obligation. Teams that prefer native Kubernetes resources can skip the Vault dialect entirely: Monopam ships a Kubernetes Operator as a Helm chart, with MonoStore and MonoSecret CRDs, and it has passed its end-to-end suite on both kind and GKE. Run both side by side; the pipeline keeps its placeholders while new workloads adopt CRDs at their own pace.

Where Monofor fits

None of this is an argument against the tooling you chose; that tooling is good, which is exactly why it should survive the move. Monopam's Vault-compatible API lets the vault change while the workflow does not, and puts every secret under one roof of grants, audit and rollback.

The hands-on walkthrough, from enabling the API to a green ExternalSecret on GKE, lives at Vault-compatible secrets; the broader platform is on the secrets management page.

Further reading: the secrets management and machine identity glossary entries.

Tagssecrets-managementvaultkubernetesdevopsmonopam

Ready to start managing
identities the right way?

Spin up a fully-loaded trial tenant in under five minutes. No credit card. No sales gate.