Repository navigation
Add read-only preflight for existing AKS Foundry hosting - #10828
Open
Yi Liu (yiliuTo) wants to merge 4 commits into
Open
Yi Liu (yiliuTo) wants to merge 4 commits into
Yi Liu (yiliuTo) wants to merge 4 commits into
Conversation
Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
Yi Liu (yiliuTo)
requested review from
a team,
JerryYang (JerryYangKai),
FanH (Siglud),
Yimin-Jin,
Eric Chen (eriolchan),
Glenn Harper (glharper),
Hui Miao (huimiu),
Zhijie Huang (hund030),
Wei Meng (m5i-work),
John Miller (therealjohn),
Travis Angevine (trangevi) and
JianW (v1212)
as code owners
November 3, 2026 14:41
|
Azure Pipelines: Successfully started running 3 pipeline(s). 24 pipeline(s) were filtered out due to trigger conditions. There may be pipelines that require an authorized user to comment /azp run to run. |
github-actions
Bot
added
ext-agents
Contributor
There was a problem hiding this comment.
🟡 Changes recommended
Critical preflight gaps let customized AKS deployments skip validation and can falsely confirm role-assignment authorization by accepting unevaluated conditions or checking the wrong scope.
5 open issues
What changed in this PR
Adds a read-only preflight to the Foundry provisioning provider’s existing-AKS deployment and ARM what-if paths, building on #10824’s supplied-resource support without changing deployment templates or teardown behavior.
Changes:
- Derives resources, grants, and operation scopes from compiled ARM templates and merged parameters.
- Adds AKS, supporting-resource, networking, and authorization checks with mandatory-failure gates.
- Adds compatibility tests, shared Bicep test utilities, Azure SDK dependencies, and documentation.
This pull request requires changes. Unresolved findings cover customized-template preflight bypasses, conditional and group-derived authorization, incorrect permission scopes, unsupported child-resource grants, evidence classification and reporting, warning persistence, production-path test coverage, unused code, and feature-stage documentation.
| File | Description |
|---|---|
docs/reference/feature-status.md |
Records existing-AKS hosting maturity and preflight support. |
docs/guides/creating-an-extension.md |
Links the read-only provisioning-provider example. |
docs/architecture/extension-framework.md |
Describes Foundry preflight integration. |
cli/azd/extensions/azure.ai.projects/README.md |
Documents checks, permissions, limitations, and remediation. |
cli/azd/extensions/azure.ai.projects/internal/testutil/bicep.go |
Centralizes local Bicep discovery and version validation. |
cli/azd/extensions/azure.ai.projects/internal/synthesis/templates_embed.go |
Clarifies preflight’s runtime Bicep usage. |
cli/azd/extensions/azure.ai.projects/internal/synthesis/schema_test.go |
Reuses shared Bicep test helpers. |
cli/azd/extensions/azure.ai.projects/internal/synthesis/hosting_test.go |
Updates Bicep-helper test calls. |
cli/azd/extensions/azure.ai.projects/internal/provisioning/onboarding_naming.go |
Evaluates constant naming functions through Bicep. |
cli/azd/extensions/azure.ai.projects/internal/provisioning/onboarding_naming_test.go |
Tests naming evaluation, failures, and cleanup. |
cli/azd/extensions/azure.ai.projects/internal/provisioning/foundry_provisioning_provider.go |
Passes progress callbacks into preflight preparation. |
cli/azd/extensions/azure.ai.projects/internal/provisioning/byo_template_plan.go |
Resolves compiled resources, nested deployments, and scopes. |
cli/azd/extensions/azure.ai.projects/internal/provisioning/byo_template_plan_test.go |
Tests nested names, scopes, and references. |
cli/azd/extensions/azure.ai.projects/internal/provisioning/byo_plan_bindings.go |
Maps resources, grants, and network operations into preflight. |
cli/azd/extensions/azure.ai.projects/internal/provisioning/aks_hosting.go |
Runs preflight and reports results before binding resolution. |
cli/azd/extensions/azure.ai.projects/internal/provisioning/aks_hosting_compatibility_test.go |
Covers binding contracts and customized-template compatibility. |
cli/azd/extensions/azure.ai.projects/internal/onboarding/types.go |
Defines plans, evidence, reports, and blocking errors. |
cli/azd/extensions/azure.ai.projects/internal/onboarding/resource_sdk.go |
Collects read-only Azure resource evidence. |
cli/azd/extensions/azure.ai.projects/internal/onboarding/resource_rules.go |
Implements AKS, AgentPool, and Storage compatibility rules. |
cli/azd/extensions/azure.ai.projects/internal/onboarding/resource_clients.go |
Shares ARM clients within a preflight pass. |
cli/azd/extensions/azure.ai.projects/internal/onboarding/plan.go |
Parses resource IDs and provides test-used plan construction. |
cli/azd/extensions/azure.ai.projects/internal/onboarding/pipeline.go |
Combines resource and authorization checks with deployment gating. |
cli/azd/extensions/azure.ai.projects/internal/onboarding/parallel_reads.go |
Bounds concurrent evidence reads. |
cli/azd/extensions/azure.ai.projects/internal/onboarding/operator_authorization.go |
Checks operator permissions and role-assignment authority. |
cli/azd/extensions/azure.ai.projects/internal/onboarding/network_sdk.go |
Reads selected subnets and parent VNets. |
cli/azd/extensions/azure.ai.projects/internal/onboarding/network_configuration.go |
Checks subnet delegation, capacity, containment, and overlap. |
cli/azd/extensions/azure.ai.projects/internal/onboarding/managed_node_network.go |
Resolves managed node subnets through VMSS evidence. |
cli/azd/extensions/azure.ai.projects/internal/onboarding/identity_authorization.go |
Evaluates supplied-identity grants and permissions. |
cli/azd/extensions/azure.ai.projects/internal/onboarding/deployment_plan.go |
Models standard-network operations and permission scopes. |
cli/azd/extensions/azure.ai.projects/internal/onboarding/authorization_test.go |
Tests authorization scopes, inheritance, caching, and failures. |
cli/azd/extensions/azure.ai.projects/internal/onboarding/authorization_sdk.go |
Coordinates authorization evidence and scoped operation checks. |
cli/azd/extensions/azure.ai.projects/internal/onboarding/authorization_rules.go |
Defines grant requirements, action matching, and caller claims. |
cli/azd/extensions/azure.ai.projects/internal/onboarding/authorization_evidence.go |
Caches authorization evidence within each pass. |
cli/azd/extensions/azure.ai.projects/go.sum |
Records added Azure SDK dependency checksums. |
cli/azd/extensions/azure.ai.projects/go.mod |
Adds Compute, Network, and Storage SDK dependencies. |
cli/azd/extensions/azure.ai.agents/README.md |
Links existing-AKS preflight documentation. |
cli/azd/extensions/azure.ai.agents/internal/synthesis/templates_embed.go |
Mirrors the Projects runtime-Bicep comment update. |
cli/azd/.vscode/cspell.yaml |
Adds file-scoped preflight terminology. |
Review effort: Max
| check.Missing = "operator object ID" | ||
| return check | ||
| } | ||
| check.Status, check.Observed, check.Cause = a.basicAction(ctx, scope, caller, roleAssignmentWrite, false) |
Comment on lines
+117
to
+120
| plan, err := p.identifyBYOTemplatePlan(ctx, source) | ||
| if err != nil { | ||
| return err | ||
| } |
Comment on lines
+309
to
+313
| if checkID == "P07" { | ||
| if grant.Missing != "" { | ||
| plan.RegistryGrantMissing = grant.Missing | ||
| } else { | ||
| plan.RegistryGrantPlanned = new(true) |
Comment on lines
+172
to
+173
| if output.Len() > 0 { | ||
| progress(output.String()) |
Comment on lines
+347
to
+350
| if node.env.scope.groupExists { | ||
| return node.env.scope.groupARMID | ||
| } | ||
| return node.env.scope.subscriptionARMID |
Preserve registry lifecycle validation, process feature gates, and parallel binding reads alongside read-only preflight. Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
Contributor
There was a problem hiding this comment.
🟡 Changes recommended
Critical preflight defects can bypass required authorization checks, block valid deployments, or falsely report sufficient permissions because identities, grants, and resource scopes are handled incorrectly.
17 open issues
Review effort: Max
Comment on lines
+170
to
+173
| token, tokenErr := a.credential.GetToken(ctx, policy.TokenRequestOptions{ | ||
| Scopes: []string{strings.TrimRight(audience, "/") + "/.default"}, | ||
| }) | ||
| caller, identityErr = identityFromToken(token.Token) |
Comment on lines
+42
to
+46
| future := grant.checkID == "G02" && !plan.Workload.Supplied && facts.absent(plan.Workload) || | ||
| grant.checkID == "G03" && !plan.Storage.Supplied && facts.absent(plan.Storage) | ||
| if future { | ||
| report = append(report, resourceResult(check, NotApplicable, | ||
| "dependency does not yet exist; the same ARM graph will create the dependency and grant")) |
| return Unverified, "operator role assignments could not be read", err | ||
| } | ||
| provenAssignments := map[*armauthorization.RoleAssignment]bool{} | ||
| if caller.groupOverage { |
Comment on lines
+105
to
+107
| if unconditionalOnly && (authString(p.Condition) != "" || authString(p.ConditionVersion) != "" || | ||
| authString(p.DelegatedManagedIdentityResourceID) != "") { | ||
| continue |
Comment on lines
+38
to
+41
| segments := 9 | ||
| if strings.EqualFold(resourceType, "Microsoft.Network/virtualNetworks/subnets") { | ||
| segments = 11 | ||
| } |
Comment on lines
+403
to
+405
| joinScope := resource.ID | ||
| if !resource.Supplied { | ||
| joinScope = vnet |
Comment on lines
+405
to
+408
| joinScope = vnet | ||
| if resource.LifecycleMissing == "" { | ||
| plan.AddOperation("P03", "Microsoft.Network/virtualNetworks/subnets/write", vnet) | ||
| } |
| return nil, err | ||
| } | ||
| if err := p.prepareBYO(ctx, src); err != nil { | ||
| if err := p.prepareBYO(ctx, src, progress); err != nil { |
Comment on lines
+34
to
+49
| // CheckClusterDeployment checks deployment-write permission in the AKS resource group without checking supporting resources. | ||
| func CheckClusterDeployment( | ||
| ctx context.Context, clusterID string, credential azcore.TokenCredential, options *arm.ClientOptions, | ||
| ) (Report, error) { | ||
| cluster, err := ParseID(clusterID, "Microsoft.ContainerService/managedClusters") | ||
| if err != nil { | ||
| return nil, err | ||
| } | ||
| a := &scopedAuthorizer{credential: credential, options: onboardingARMOptions(options)} | ||
| a.reads = newAuthorizationEvidence(a.credential, a.options) | ||
| caller, tokenErr, identityErr := a.operatorIdentity(ctx) | ||
| report := a.checkOperations(ctx, []operation{ | ||
| {"P09", "Microsoft.Resources/deployments/write", cluster.Parent.String()}, | ||
| }, caller, tokenErr, identityErr) | ||
| return report, report.Err() | ||
| } |
| | Azure Functions | Stable | | ||
| | Azure Kubernetes Service (AKS) | Beta | | ||
| | Azure AI | Beta | | ||
| | [Microsoft Foundry on existing AKS](../../cli/azd/extensions/azure.ai.projects/README.md#existing-aks-hosting) (default or supplied supporting resources, read-only preflight) | In development (infrastructure enabled by default; Agents init/adoption disabled by default) | |
Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
Contributor
There was a problem hiding this comment.
🔵 Needs a closer look
Active AKS hosting can bypass preflight when its parameter is disabled or absent, and P07 can falsely approve role assignments despite assignment restrictions or insufficient permissions at the actual scope.
16 open issues
1 resolved since last review
Review effort: Max
Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
Contributor
There was a problem hiding this comment.
🔵 Needs a closer look
Templates with hard-coded AKS hosting or a differently named parameter can bypass AKS, resource, and authorization preflight checks in both deployment and preview.
16 open issues
Review effort: Max
Contributor
azure.ai.agents PR buildNote This is an unsigned development build. Install it only if you trust this PR. Install the extension: azd ext install "https://azuresdkartifacts.z5.web.core.windows.net/azd/extensions/pr/10828/azure-ai-agents.zip"
|
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Validation
azure.ai.projectsandazure.ai.agents.go test ./... -count=1. The color-sensitive command tests requireNO_COLORto be unset at process startup.Infra|BYOAKS|BYOAks.Related: #10824.