feat(compute): delegate sandbox authentication to drivers

Signed-off-by: Drew Newberry <385+drew@users.noreply.github.com>

Author: drew

.agents/skills/debug-openshell-cluster/SKILL.md

@@ -460,7 +460,10 @@ Then inspect sandbox resources in that namespace.
 Check the configured sandbox service account when TokenReview bootstrap or
 sandbox registration fails. Helm creates a dedicated sandbox service account by
 default and writes it to `[openshell.drivers.kubernetes].service_account_name`;
-the gateway rejects projected tokens from other service accounts.
+the selected Kubernetes compute driver rejects projected tokens from other
+service accounts. For an external driver, inspect its logs and confirm it
+advertises `supports_sandbox_authentication`; the gateway delegates the opaque
+credential over the driver socket and never interprets Kubernetes settings.
 
 ```bash
 helm -n openshell get values openshell | grep -A3 sandboxServiceAccount

.agents/skills/helm-dev-environment/SKILL.md

@@ -217,7 +217,9 @@ annotations to `spiffe://openshell.local/openshell/sandbox/<sandbox-id>`.
 OpenShell mounts the SPIFFE CSI Workload API socket at
 `/spiffe-workload-api/spire-agent.sock` into sandbox pods for provider token
 grants. Supervisor-to-gateway authentication remains on the Kubernetes
-ServiceAccount bootstrap and gateway-minted sandbox JWT path.
+ServiceAccount bootstrap and gateway-minted sandbox JWT path; the selected
+Kubernetes compute driver validates the projected token before the gateway
+mints its JWT.
 
 ---
 

architecture/compute-runtimes.md

@@ -169,6 +169,15 @@ The driver reports this behavior through
 in-process and external drivers. Older drivers omit the field and retain the
 conservative operator-managed behavior.
 
+Drivers that can verify a platform-native sandbox credential advertise
+`GetCapabilities.supports_sandbox_authentication`. On the path-scoped
+`IssueSandboxToken` exchange, the gateway forwards the opaque bearer credential
+to that selected driver through `AuthenticateSandbox`. The driver returns only
+the authenticated sandbox ID. The gateway then verifies that its durable
+sandbox record exists and mints the gateway JWT. The driver socket is therefore
+a sandbox-identity trust boundary, but it does not grant user or administrator
+authority.
+
 ## Deletion Lifecycle
 
 Lifecycle requests use per-sandbox gates to serialize stop, start, and
@@ -457,20 +466,23 @@ watcher emits only sandbox CR changes, not platform events.
 
 ### SA Token Authentication
 
-The gateway's `K8sServiceAccountAuthenticator` adapts its `NamespaceValidator`
-per mode (`crates/openshell-server/src/auth/k8s_sa.rs`):
+The Kubernetes driver's `AuthenticateSandbox` implementation applies its named
+`[openshell.drivers.kubernetes]` configuration per mode:
 
 - **Shared:** `Exact` — accepts only the single configured namespace.
 - **Managed:** `Prefix` — accepts any namespace starting with `openshell-{gateway_id}-`.
 - **Operator:** `Allowlist` — accepts namespaces present in the dynamic
   `BTreeSet` populated by the label/file watchers. Starts empty (fail-closed)
   until the first watcher update.
 
-These checks rely on an ownership invariant. In shared and managed modes, the
-gateway and its trusted Agent Sandbox controller exclusively administer the
-sandbox namespace, Sandbox CRs, sandbox pods, and configured sandbox
-ServiceAccount. Other principals must not create or mutate those resources or
-use that ServiceAccount. In operator mode, the platform operator retains
+It validates the projected token with Kubernetes `TokenReview`, checks the live
+pod UID, and verifies the pod's controlling Sandbox CR UID and sandbox ID before
+returning the identity to the gateway. These checks rely on an ownership
+invariant. In shared and managed modes, the Kubernetes driver and its trusted
+Agent Sandbox controller exclusively administer the sandbox namespace, Sandbox
+CRs, sandbox pods, and configured sandbox ServiceAccount. Other principals must
+not create or mutate those resources or use that ServiceAccount. In operator
+mode, the platform operator retains
 namespace lifecycle ownership, but must preserve the same exclusive control of
 Sandbox CRs and the pods and ServiceAccount used for sandbox token bootstrap.
 An allowlisted namespace is therefore a trust grant, not a tenant isolation

architecture/gateway.md

@@ -225,11 +225,12 @@ identity inspection without client-side token decoding.
 Sandbox secrets are gateway-signed JWTs bound to a single sandbox ID. Docker,
 Podman, and VM drivers deliver the initial token through supervisor-only
 runtime material; Kubernetes supervisors exchange a projected ServiceAccount
-token through `IssueSandboxToken`. The gateway validates that projected token
-with Kubernetes `TokenReview`, requires the configured sandbox service account,
-checks the returned pod binding against the live pod UID, and verifies the pod's
-controlling `Sandbox` ownerReference against the live Sandbox CR UID and
-sandbox-id label before minting the gateway JWT. The bootstrap path accepts
+token through `IssueSandboxToken`. The gateway delegates that opaque credential
+to the selected compute driver's `AuthenticateSandbox` RPC. A capable driver is
+trusted to return the authenticated sandbox ID, while the gateway still requires
+a matching durable sandbox record before minting a JWT. The Kubernetes driver
+uses its own named configuration to run TokenReview and verify the live pod and
+controlling Sandbox CR. The bootstrap path accepts
 both `agents.x-k8s.io/v1beta1` ownerReferences from newer Agent Sandbox
 controllers and `agents.x-k8s.io/v1alpha1` ownerReferences from existing
 deployments. Supervisors renew gateway JWTs in memory before expiry only while

crates/openshell-driver-docker/src/lib.rs

@@ -635,6 +635,7 @@ impl DockerComputeDriver {
             driver_version: self.config.daemon_version.clone(),
             default_image: self.config.default_image.clone(),
             gateway_manages_lifecycle: true,
+            supports_sandbox_authentication: false,
         }
     }
 
@@ -1711,6 +1712,16 @@ impl DockerComputeDriver {
 
 #[tonic::async_trait]
 impl ComputeDriver for ComputeDriverService {
+    async fn authenticate_sandbox(
+        &self,
+        _request: Request<openshell_core::proto::compute::v1::AuthenticateSandboxRequest>,
+    ) -> Result<Response<openshell_core::proto::compute::v1::AuthenticateSandboxResponse>, Status>
+    {
+        Err(Status::unimplemented(
+            "docker does not authenticate sandbox credentials",
+        ))
+    }
+
     type WatchSandboxesStream = WatchStream;
 
     async fn get_capabilities(
@@ -1873,6 +1884,16 @@ impl ComputeDriver for ComputeDriverService {
 
 #[tonic::async_trait]
 impl ComputeDriver for DockerComputeDriver {
+    async fn authenticate_sandbox(
+        &self,
+        _request: Request<openshell_core::proto::compute::v1::AuthenticateSandboxRequest>,
+    ) -> Result<Response<openshell_core::proto::compute::v1::AuthenticateSandboxResponse>, Status>
+    {
+        Err(Status::unimplemented(
+            "docker does not authenticate sandbox credentials",
+        ))
+    }
+
     type WatchSandboxesStream = WatchStream;
 
     async fn get_capabilities(

crates/openshell-driver-kubernetes/README.md

@@ -20,7 +20,7 @@ workspace namespace modes via `workspace_mode`:
   Kubernetes namespace.
 
 Workspace namespace modes assume exclusive control of the sandbox identity
-resource chain. In shared and managed modes, only the gateway and its trusted
+resource chain. In shared and managed modes, only the driver and its trusted
 Agent Sandbox controller may administer the sandbox namespace, Sandbox CRs,
 sandbox pods, or configured sandbox ServiceAccount. In operator mode, the
 platform operator owns namespace lifecycle but must prevent other principals
@@ -88,7 +88,9 @@ Sandbox pods run as `service_account_name` and keep
 `automountServiceAccountToken: false`. The only Kubernetes token exposed to the
 supervisor is an explicit, audience-bound projected token mounted at
 `/var/run/secrets/openshell/token` for the one-shot `IssueSandboxToken`
-bootstrap exchange.
+bootstrap exchange. The Kubernetes driver authenticates that token through the
+compute-driver protocol using its own `service_account_name` and workspace-mode
+namespace policy; the gateway receives only the verified sandbox ID.
 
 The gateway uses the supervisor relay for connect, exec, and file sync. Sandbox
 pods do not need direct external ingress for SSH.