Add Azure.Mcp.Tools.SreAgent (scaffolding for issue #2541)

[BandaruDheeraj]

Summary

New tool group Azure.Mcp.Tools.SreAgent (resolves #2541) that ports the SRE Agent MCP server from the Node implementation in microsoft/sreagent-runtime to the .NET azmcp server.

The tool group exposes 57 commands across 11 sub-groups for managing and interacting with Azure SRE Agent (Microsoft.App/SREAgentPreview) resources and their data-plane API (https://*.azuresre.ai).

Scope

Sub-groups + commands

Sub-group	Commands
agents	list get create delete
agents tools	list get create
skills	list create delete
connectors	list get create-kusto create-mcp delete test
hooks	list get delete
hooks thread	list activate deactivate
threads	list get create send-message delete investigate investigate-yolo
scheduledtasks	list get create delete pause resume
incidents	active-list create plans-list plans-create setup-pagerduty setup-servicenow
workflows	generate validate apply
docs	get plus memories-list memories-search memories-add memories-delete memories-reindex
architecture	plan
commonprompts	list get create delete

Implementation

• ARM enumeration of Microsoft.App/SREAgentPreview resources via Azure Resource Graph (BaseAzureResourceService.ExecuteResourceQueryAsync<T>).
• Data-plane calls use a bearer token for https://azuresre.dev/.default (matches the audience the existing .NET CLI in the SRE Agent runtime repo uses).
• SreAgentService injects IHttpClientFactory per the repo convention (unblocks recorded-test migration).
• Envelope shapes, field names, and error semantics mirror the Node MCP server at HEAD ce366bf0a3 so existing SRE Agent dashboards/clients see the same JSON.
• CommandMetadata populated for every command (Id, Title, Description, Destructive, Idempotent, OpenWorld, ReadOnly, Secret, LocalRequired).

Tests

114 unit tests in Azure.Mcp.Tools.SreAgent.UnitTests (6 existing + 108 new). Every ported command has a Constructor_InitializesCommandCorrectly and RegisterOptions_AddsExpectedOptions test backed by the shared CommandUnitTestsBase<TCommand, TService> harness. Live end-to-end testing against an SRE Agent resource has not been performed in this PR.

Changelog

Adds a Features Added entry under Azure.Mcp.Server describing the new SRE Agent tool group.

Files

• tools/Azure.Mcp.Tools.SreAgent/src - service, commands, models, options, setup
• tools/Azure.Mcp.Tools.SreAgent/tests - unit tests
• servers/Azure.Mcp.Server/src/Areas/AreaSetupFactory.cs - registration
• servers/Azure.Mcp.Server/CHANGELOG.md - feature entry

Issue

Closes #2541

[BandaruDheeraj]

@g2vinay this is now ready for review.

Summary:

• New tool group: Azure.Mcp.Tools.SreAgent (issue [ONBOARD] Azure SRE Agent Tools #2541)
• ~48 commands across 11 sub-groups: agents, skills, connectors, hooks, threads, scheduledtasks, incidents, workflows, docs, architecture, memories
• ARM enumeration of Microsoft.App/SREAgentPreview via Resource Graph + data-plane HTTP via CallDataPlaneAsync (token audience https://azuresre.dev/.default)
• New UnitTests project (xunit.v3 + NSubstitute) — all green
• Added to Microsoft.Mcp.slnx, servers/Azure.Mcp.Server/README.md, and changelog-entries/
• Full Azure.Mcp.Server build green

Implementation was fanned out across four sub-agent branches (A: agents+skills, B: connectors+hooks, C: threads+scheduledtasks, D: incidents+workflows+docs+architecture) and integrated via #region (sub-agent X) block markers. Happy to address any feedback or split into smaller PRs if preferred.

[Copilot]

Pull request overview

This PR introduces a new Azure MCP tool package, Azure.Mcp.Tools.SreAgent, wiring it into the Azure MCP Server and adding initial command surface area (agents/sub-agents, tools/skills, connectors, hooks, threads, scheduled tasks, incidents, workflows, docs, and architecture planning).

Changes:

• Added the Azure.Mcp.Tools.SreAgent tool project with commands, options, models, JSON source-gen context, and an ISreAgentService + implementation for ARM discovery and SRE Agent data-plane calls.
• Registered the new sreagent area in Azure.Mcp.Server and updated solution (.slnx) entries accordingly.
• Added server README examples and a changelog entry, plus a starter unit test project and tests for agents list.

Invoking Livetests

Copilot submitted PRs are not trustworthy by default. Users with write access to the repo need to validate the contents of this PR before leaving a comment with the text /azp run mcp - pullrequest - live. This will trigger the necessary livetest workflows to complete required validation.

Reviewed changes

Copilot reviewed 126 out of 126 changed files in this pull request and generated 15 comments.

Show a summary per file

File	Description
tools/Azure.Mcp.Tools.SreAgent/src/Azure.Mcp.Tools.SreAgent.csproj	New tool project (AOT-compatible) and dependencies.
tools/Azure.Mcp.Tools.SreAgent/src/AssemblyInfo.cs	InternalsVisibleTo for unit/live tests.
tools/Azure.Mcp.Tools.SreAgent/src/GlobalUsings.cs	Adds global usings for the tool project.
tools/Azure.Mcp.Tools.SreAgent/src/SreAgentSetup.cs	Registers DI services and builds the sreagent command tree.
tools/Azure.Mcp.Tools.SreAgent/src/Services/ISreAgentService.cs	Defines the service contract for ARM + data-plane operations.
tools/Azure.Mcp.Tools.SreAgent/src/Services/SreAgentService.cs	Implements ARM discovery + data-plane REST calls.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/BaseSreAgentCommand.cs	Base subscription-scoped command wiring (adds RG option binding).
tools/Azure.Mcp.Tools.SreAgent/src/Commands/SreAgentDataPlaneCommand.cs	Base for data-plane commands (binds --agent and resolves endpoint).
tools/Azure.Mcp.Tools.SreAgent/src/Commands/SreAgentCommandHelpers.cs	Shared helpers for endpoint resolution and JSON parsing.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/SreAgentPortedCommandHelpers.cs	Helper utilities for text results and JSON array extraction.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/SreAgentJsonContext.cs	System.Text.Json source-generation context for tool models/results.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Agents/AgentsListCommand.cs	Lists SRE Agent ARM resources in a subscription/RG.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Agents/AgentsGetCommand.cs	Gets a sub-agent definition from an SRE Agent resource.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Agents/AgentsCreateCommand.cs	Creates/updates a sub-agent on a targeted agent resource.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Agents/AgentsDeleteCommand.cs	Deletes a sub-agent (requires confirm).
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Agents/AgentsToolsGetCommand.cs	Gets a custom tool definition from a targeted agent resource.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Agents/AgentsToolsCreateCommand.cs	Creates/updates a custom tool definition on an agent resource.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Skills/AgentToolsListCommand.cs	Lists custom tools on a targeted agent resource.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Skills/SkillsListCommand.cs	Lists custom skills on a targeted agent resource.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Skills/SkillsCreateCommand.cs	Creates/updates a skill on a targeted agent resource.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Skills/SkillsDeleteCommand.cs	Deletes a tool/skill (requires confirm).
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Connectors/ConnectorsListCommand.cs	Lists connectors on a targeted agent resource.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Connectors/ConnectorsGetCommand.cs	Gets a connector’s details.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Connectors/ConnectorsCreateKustoCommand.cs	Creates/updates a Kusto connector.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Connectors/ConnectorsCreateMcpCommand.cs	Creates/updates an MCP connector (stdio/http).
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Connectors/ConnectorsDeleteCommand.cs	Deletes a connector.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Connectors/ConnectorsTestCommand.cs	Tests a connector and returns exposed tools.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Hooks/HooksListCommand.cs	Lists hooks on a targeted agent resource.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Hooks/HooksGetCommand.cs	Gets hook details.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Hooks/HooksDeleteCommand.cs	Deletes a hook.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Hooks/HooksThreadListCommand.cs	Lists hook activation state for a thread.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Hooks/HooksThreadActivateCommand.cs	Activates a thread hook.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Hooks/HooksThreadDeactivateCommand.cs	Deactivates a thread hook.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Threads/ThreadsCommandBase.cs	Shared thread polling/follow-up classification helpers.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Threads/ThreadsListCommand.cs	Lists SRE Agent threads.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Threads/ThreadsGetCommand.cs	Gets thread messages.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Threads/ThreadsCreateCommand.cs	Creates a thread and waits for initial completion.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Threads/ThreadsSendMessageCommand.cs	Sends a message and polls for completion.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Threads/ThreadsDeleteCommand.cs	Deletes a thread (requires confirm).
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Threads/ThreadsInvestigateCommand.cs	Starts an investigation loop with follow-up handling.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Threads/ThreadsInvestigateYoloCommand.cs	YOLO variant that auto-approves pending approvals.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/ScheduledTasks/ScheduledTasksListCommand.cs	Lists scheduled tasks.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/ScheduledTasks/ScheduledTasksGetCommand.cs	Gets a scheduled task.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/ScheduledTasks/ScheduledTasksCreateCommand.cs	Creates a scheduled task.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/ScheduledTasks/ScheduledTasksDeleteCommand.cs	Deletes a scheduled task (requires confirm).
tools/Azure.Mcp.Tools.SreAgent/src/Commands/ScheduledTasks/ScheduledTasksPauseCommand.cs	Pauses a scheduled task.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/ScheduledTasks/ScheduledTasksResumeCommand.cs	Resumes a scheduled task.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Incidents/IncidentsPlansListCommand.cs	Lists incident response plans (filters/handlers).
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Incidents/IncidentsPlansCreateCommand.cs	Creates and enables an incident response plan.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Incidents/IncidentsActiveListCommand.cs	Lists “incident-like” active threads.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Incidents/IncidentsCreateCommand.cs	Creates an incident investigation thread.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Incidents/IncidentsSetupPagerdutyCommand.cs	Creates PagerDuty connector from env var credentials.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Incidents/IncidentsSetupServicenowCommand.cs	Creates ServiceNow connector from env var credentials.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Workflows/WorkflowsGenerateCommand.cs	Generates agent/tool YAML scaffolding.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Workflows/WorkflowsValidateCommand.cs	Validates YAML content for common issues.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Workflows/WorkflowsApplyCommand.cs	Applies YAML to the data-plane API via minimal parsing.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Docs/DocsGetCommand.cs	Returns SRE Agent reference docs by topic.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Docs/MemoriesListCommand.cs	Lists knowledge base documents.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Docs/MemoriesSearchCommand.cs	Searches knowledge base documents.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Docs/MemoriesAddCommand.cs	Uploads markdown to knowledge base.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Docs/MemoriesDeleteCommand.cs	Deletes a knowledge base document (requires confirm).
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Docs/MemoriesReindexCommand.cs	Triggers knowledge base reindex.
tools/Azure.Mcp.Tools.SreAgent/src/Commands/Architecture/PlanCommand.cs	Produces an architecture plan based on requirements.
tools/Azure.Mcp.Tools.SreAgent/src/Options/BaseSreAgentOptions.cs	Base options including agent + subscription options.
tools/Azure.Mcp.Tools.SreAgent/src/Options/SreAgentOptionDefinitions.cs	Option definitions for the command line surface.
tools/Azure.Mcp.Tools.SreAgent/src/Options/SreAgentPortedOptionDefinitions.cs	Additional option definitions for “ported” command surface.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Agents/AgentsListOptions.cs	Options for agents list.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Agents/AgentsGetOptions.cs	Options for agents get.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Agents/AgentsCreateOptions.cs	Options for agents create.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Agents/AgentsDeleteOptions.cs	Options for agents delete.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Agents/AgentsToolsGetOptions.cs	Options for tools get.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Agents/AgentsToolsCreateOptions.cs	Options for tools create.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Skills/AgentToolsListOptions.cs	Options for tools list.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Skills/SkillsListOptions.cs	Options for skills list.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Skills/SkillsCreateOptions.cs	Options for skills create.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Skills/SkillsDeleteOptions.cs	Options for skills delete.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Connectors/ConnectorsListOptions.cs	Options for connectors list.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Connectors/ConnectorsGetOptions.cs	Options for connectors get.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Connectors/ConnectorsDeleteOptions.cs	Options for connectors delete.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Connectors/ConnectorsTestOptions.cs	Options for connectors test.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Connectors/ConnectorsCreateKustoOptions.cs	Options for connectors create-kusto.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Connectors/ConnectorsCreateMcpOptions.cs	Options for connectors create-mcp.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Hooks/HooksListOptions.cs	Options for hooks list.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Hooks/HooksGetOptions.cs	Options for hooks get.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Hooks/HooksDeleteOptions.cs	Options for hooks delete.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Hooks/HooksThreadListOptions.cs	Options for thread hook list.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Hooks/HooksThreadActivateOptions.cs	Options for thread hook activate.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Hooks/HooksThreadDeactivateOptions.cs	Options for thread hook deactivate.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Threads/ThreadsOptions.cs	Options for thread operations and investigations.
tools/Azure.Mcp.Tools.SreAgent/src/Options/ScheduledTasks/ScheduledTasksOptions.cs	Options for scheduled task operations.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Incidents/IncidentOptions.cs	Options for incident operations and connector setup.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Docs/DocsOptions.cs	Options for docs and knowledge base operations.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Workflows/WorkflowOptions.cs	Options for workflows generate/validate/apply.
tools/Azure.Mcp.Tools.SreAgent/src/Options/Architecture/PlanOptions.cs	Options for architecture planning.
tools/Azure.Mcp.Tools.SreAgent/src/Models/SreAgentResource.cs	Lightweight ARM projection model for SRE Agent resources.
tools/Azure.Mcp.Tools.SreAgent/src/Models/SreAgentDeleteResult.cs	Delete result model.
tools/Azure.Mcp.Tools.SreAgent/src/Models/SreSubAgent.cs	Sub-agent model.
tools/Azure.Mcp.Tools.SreAgent/src/Models/SreSubAgentProperties.cs	Sub-agent properties model (extension data supported).
tools/Azure.Mcp.Tools.SreAgent/src/Models/SreSubAgentCreateRequest.cs	Sub-agent create request model.
tools/Azure.Mcp.Tools.SreAgent/src/Models/SreAgentTool.cs	Custom tool model.
tools/Azure.Mcp.Tools.SreAgent/src/Models/SreAgentToolProperties.cs	Custom tool properties model.
tools/Azure.Mcp.Tools.SreAgent/src/Models/SreAgentToolParameter.cs	Tool parameter model.
tools/Azure.Mcp.Tools.SreAgent/src/Models/SreAgentToolCreateRequest.cs	Tool create request model.
tools/Azure.Mcp.Tools.SreAgent/src/Models/SreSkill.cs	Skill model.
tools/Azure.Mcp.Tools.SreAgent/src/Models/SreSkillProperties.cs	Skill properties model (extension data supported).
tools/Azure.Mcp.Tools.SreAgent/src/Models/SreSkillCreateRequest.cs	Skill create request model.
tools/Azure.Mcp.Tools.SreAgent/src/Models/AgentConnector.cs	Connector model (incl. extended properties).
tools/Azure.Mcp.Tools.SreAgent/src/Models/AgentConnectorEnvelope.cs	Connector envelope model.
tools/Azure.Mcp.Tools.SreAgent/src/Models/ConnectorTestResult.cs	Connector test result model.
tools/Azure.Mcp.Tools.SreAgent/src/Models/ConnectorToolInfo.cs	Connector tool info model.
tools/Azure.Mcp.Tools.SreAgent/src/Models/HookEnvelope.cs	Hook envelope model.
tools/Azure.Mcp.Tools.SreAgent/src/Models/HookSpec.cs	Hook spec/properties model.
tools/Azure.Mcp.Tools.SreAgent/src/Models/HookDefinition.cs	Hook definition model.
tools/Azure.Mcp.Tools.SreAgent/src/Models/ThreadHookInfo.cs	Thread hook info model.
tools/Azure.Mcp.Tools.SreAgent/src/Models/ThreadHooksResponse.cs	Thread hooks response model.
tools/Azure.Mcp.Tools.SreAgent/src/Models/SreAgentThreadModels.cs	Thread/message/approval/user-question models.
tools/Azure.Mcp.Tools.SreAgent/src/Models/SreAgentScheduledTaskModels.cs	Scheduled task models.
tools/Azure.Mcp.Tools.SreAgent/src/Models/SreAgentPortedModels.cs	Additional models for incident/docs/memory surfaces.
tools/Azure.Mcp.Tools.SreAgent/tests/Azure.Mcp.Tools.SreAgent.UnitTests/Azure.Mcp.Tools.SreAgent.UnitTests.csproj	New unit test project for SreAgent commands.
tools/Azure.Mcp.Tools.SreAgent/tests/Azure.Mcp.Tools.SreAgent.UnitTests/AssemblyAttributes.cs	Test assembly-wide attributes.
tools/Azure.Mcp.Tools.SreAgent/tests/Azure.Mcp.Tools.SreAgent.UnitTests/Agents/AgentsListCommandTests.cs	Unit tests for agents list command behavior.
servers/Azure.Mcp.Server/src/Program.cs	Registers SreAgentSetup in server area registration.
servers/Azure.Mcp.Server/README.md	Adds SRE Agent example prompts and service list entry.
servers/Azure.Mcp.Server/changelog-entries/bandarud-sreagent-tool.yml	Adds changelog entry for the new SRE Agent tool group.
servers/Azure.Mcp.Server/Azure.Mcp.Server.slnx	Adds SreAgent project folder and project reference.
Microsoft.Mcp.slnx	Adds SreAgent project + unit test project to root solution.

[jongio]

A few items beyond the existing feedback:

HttpClient lifecycle - SreAgentService creates HTTP clients via TenantService.GetClient() (lines 101 and 714) instead of injecting IHttpClientFactory. The repo convention is to use IHttpClientFactory.CreateClient() for proper socket lifecycle management and recorded test support. This will block the recorded-test migration path.

Test coverage - Only AgentsListCommandTests exists. With ~48 commands across 11 subgroups, most commands have zero test coverage. Even basic Constructor_InitializesCommandCorrectly and BindOptions_BindsOptionsCorrectly tests per command would catch registration and option-binding bugs early.

Duplicate endpoint resolution - Both SreAgentDataPlaneCommand.ResolveEndpointAsync and SreAgentCommandHelpers.ResolveAgentEndpointAsync resolve an agent name to its data-plane endpoint URL. Pick one location and remove the other.

Formatting density - Several ported commands (e.g. IncidentsActiveListCommand, WorkflowsApplyCommand) cram entire method bodies onto single lines or collapse multi-statement blocks. This makes them very hard to review and maintain. Breaking these into standard multi-line formatting would help reviewers and future contributors.

[jongio]

This creates an HttpClient directly instead of going through IHttpClientFactory. The repo convention (per recorded test docs) is to inject IHttpClientFactory into services and call CreateClient() so the test infrastructure can intercept and record/playback HTTP traffic. Same issue on line 714 with CallAgentDataPlaneAsync.

[jongio]

ResolveEndpointAsync here duplicates SreAgentCommandHelpers.ResolveAgentEndpointAsync. Both call sreAgentService.GetAgentAsync and pull .Endpoint. Having two copies means a fix in one won't propagate. Consider removing one and calling the other.

[BandaruDheeraj]

Security review complete — ready for review

Ran a focused security review on this PR and addressed all High and Medium findings in commit a9875999.

Fixes applied

High severity

1. Connector secret redaction — ToConnector (read path) and the CreateOrUpdateConnectorAsync fallback (write path) now redact a unified set of sensitive keys before returning the connector to the LLM/caller: bearerToken, apiKey, password, secret, clientSecret, accessToken, refreshToken, token, authorization. Case-insensitive match; also redacts nested headers.Authorization.
2. Upstream error body sanitization — SanitizeForErrorMessage strips JSON-form secrets ("<key>":"<value>") and raw Authorization: Bearer/Basic/Token/ApiKey ... headers from upstream response bodies before they're surfaced in thrown exceptions / tool errors.

Medium severity
3. HTTPS enforcement on memory upload — UploadMemoryAsync now rejects non-https:// server URLs.
4. ServiceNow instance allowlist — extended ValidateServiceNowInstanceUrl to accept both .service-now.com and .servicenowservices.com (the latter is used by some managed/government tenants).

Validation

• 138/138 unit tests passing (incl. new regression tests for the redaction + sanitization paths).
• Live-validated against post-ga-dbandaru-sample-demo: verified agent_tools_create, agent_skills_create, and agents_create all work end-to-end against the agent runtime with the fixes in place; verified secret fields stay redacted in agent_connectors_get responses.

Node MCP parity

For completeness, the same four fixes have been mirrored into the Node MCP server (separate repo, separate PR) so both implementations stay aligned.

Ready for review — please take a look.

[g2vinay]

Left my comments below.

[jongio]

Addresses my previous feedback cleanly. The 19 new commits fix everything I flagged:

• IHttpClientFactory injected properly (_httpClientFactory.CreateClient())
• Endpoint resolver deduplicated to SreAgentCommandHelpers (DataPlaneCommand delegates)
• Per-command Constructor/BindOptions tests added across all subgroups
• Dense single-line command bodies reformatted
• All previously missing subgroup registrations (hooks, scheduledtasks, incidents, workflows, docs, architecture) now wired up
• Secret redaction added with solid test coverage
• New CommonPrompts command group added

CI is green across all platforms including native AOT builds. No new issues spotted in the incremental diff.

[g2vinay]

This PR is missing Live Tests

BLK-01 · No Live Tests — Azure Service Commands Require Them

Per new-command.md and AGENTS.md, all commands that interact with Azure resources MUST include live tests recorded for playback using RecordedCommandTestsBase. No LiveTests project exists in this PR.

The author's Round 4 note acknowledges this is deferred. Per repo policy, this is a mandatory requirement — not optional.

Required:

1. Create tools/Azure.Mcp.Tools.SreAgent/tests/Azure.Mcp.Tools.SreAgent.LiveTests/Azure.Mcp.Tools.SreAgent.LiveTests.csproj
2. Implement recorded live tests using RecordedCommandTestsBase for commands that make network calls
3. Add recordings per /docs/recorded-tests.md
4. Add the LiveTests project to Microsoft.Mcp.slnx and Azure.Mcp.Server.slnx

[g2vinay]

BLK-02 · No Live Test Infrastructure (Bicep + Post-Deployment Script)

Per new-command.md, Azure service toolsets MUST include:

• tools/Azure.Mcp.Tools.SreAgent/tests/test-resources.bicep
• tools/Azure.Mcp.Tools.SreAgent/tests/test-resources-post.ps1

Neither exists. The Bicep template must provision a Microsoft.App/agents resource with appropriate RBAC role assignments for the test application.

[g2vinay]

BLK-03 · No assets.json for Recorded Tests

Per AGENTS.md and docs/recorded-tests.md, an assets.json is required alongside the live test project:

{
  "AssetsRepo": "Azure/azure-sdk-assets",
  "AssetsRepoPrefixPath": "",
  "TagPrefix": "Azure.Mcp.Tools.SreAgent.LiveTests",
  "Tag": ""
}

[g2vinay]

BLK-04 · Multiple Options Classes per File

Per AGENTS.md: "Always put new classes and interfaces in separate files."

The following files each contain multiple classes — each class must be moved to its own file:

File	Classes (must be split)
Options/CommonPrompts/CommonPromptOptions.cs	CommonPromptsListOptions, CommonPromptsGetOptions, CommonPromptsCreateOptions, CommonPromptsDeleteOptions
Options/ScheduledTasks/ScheduledTasksOptions.cs	ScheduledTasksListOptions, ScheduledTasksGetOptions, ScheduledTasksCreateOptions, ScheduledTasksDeleteOptions, ScheduledTasksPauseOptions, ScheduledTasksResumeOptions
Options/Incidents/IncidentOptions.cs	IncidentPlanCreateOptions, IncidentRemoteOptions, IncidentConnectorPagerDutyOptions, IncidentConnectorServiceNowOptions, IncidentCreateOptions
Options/Threads/ThreadsOptions.cs	ThreadsListOptions, ThreadsGetOptions, ThreadsCreateOptions, ThreadsSendMessageOptions, ThreadsDeleteOptions, ThreadsInvestigateOptions, ThreadsInvestigateYoloOptions
Options/Docs/DocsOptions.cs	DocsGetOptions, MemoryRemoteOptions, MemoriesSearchOptions, MemoriesAddOptions, MemoriesDeleteOptions
Options/Workflows/WorkflowOptions.cs	WorkflowsGenerateOptions, WorkflowsValidateOptions, WorkflowsApplyOptions

[g2vinay]

BLK-05 · azmcp-commands.md and e2eTestPrompts.md Not Updated

Per AGENTS.md:

"Update /servers/Azure.Mcp.Server/docs/azmcp-commands.md with new command details"
"Add test prompts to /servers/Azure.Mcp.Server/docs/e2eTestPrompts.md (maintain alphabetical order)"

README.md was updated with example prompts, but the authoritative command reference and test prompts files were not updated for any of the 57 new commands. These are used by AI agents for tool discovery.

[g2vinay]

BLK-08 · CODEOWNERS Not Updated

Per AGENTS.md: "Add CODEOWNERS entry for new toolset." No CODEOWNERS entry was added.

[g2vinay]

HIGH-01 · SreAgentSetup Is Not Sealed

Per AGENTS.md: "Make command classes sealed unless designed for inheritance."

// Current
public class SreAgentSetup : IAreaSetup

// Required
public sealed class SreAgentSetup : IAreaSetup

[g2vinay]

HIGH-02 · "Ported" Delete Commands Return HTTP 200 for User Errors

CommonPromptsDeleteCommand (and potentially other ported commands) returns a text error with SetTextResult when --confirm is missing — this produces HTTP 200 with error text in the body.

// Wrong — HTTP 200 with error text
SreAgentPortedCommandHelpers.SetTextResult(context.Response,
    $"Error: Refusing to delete common prompt '{o.Name}': ...");
return context.Response;

The correct pattern (already used by AgentsDeleteCommand and ConnectorsDeleteCommand) is:

// Correct — maps to HTTP 400
throw new InvalidOperationException($"Refusing to delete ...: destructive operation requires --confirm true.");

All "ported" commands that use SetTextResult to report user errors must be updated to throw instead.

[g2vinay]

HIGH-04 · Duplicate Option Definitions Across Two Static Classes

SreAgentOptionDefinitions.cs and SreAgentPortedOptionDefinitions.cs both define: Agent, Name, Description, Confirm, AuthType, Database, Query, UrlTemplate, Parameters, Connector, Handoffs, Tools. This causes confusion about which definition to use and creates divergence risk. Consolidate into a single SreAgentOptionDefinitions.cs with static nested classes per sub-group if needed.

[g2vinay]

HIGH-06 · SSRF Risk — agentEndpoint from ARM Not Validated Against Known Domain

CallDataPlaneAsync validates HTTPS but does not validate the endpoint host. The endpoint is read directly from properties.agentEndpoint in the ARM response. A compromised subscription could set this to an attacker-controlled HTTPS domain, causing the server to forward bearer tokens there.

Fix — validate before acquiring token:

private static void ValidateDataPlaneEndpoint(Uri endpointUri)
{
    if (!endpointUri.Host.EndsWith(".azuresre.ai", StringComparison.OrdinalIgnoreCase))
    {
        throw new ArgumentException(
            $"SRE Agent endpoint host '{endpointUri.Host}' is not a recognized *.azuresre.ai domain.",
            nameof(endpointUri));
    }
}

Call this in CallDataPlaneAsync before GetTokenAsync.

[g2vinay]

HIGH-07 · PauseScheduledTaskAsync / ResumeScheduledTaskAsync Pass Null Body to POST

await CallDataPlaneAsync(endpoint, $"/api/v1/scheduledtasks/{taskId}/pause", HttpMethod.Post,
    tenant: tenant, cancellationToken: cancellationToken);  // ← jsonBody omitted (null)

ActivateThreadHookAsync passes "{}" for the same type of action-POST. If the SRE Agent API requires a non-null body for POST operations these calls will fail with 400 or 415. Pass "{}" as jsonBody for consistency.

---

[g2vinay]

HIGH-08 · AgentToolsListCommand in Wrong Folder / Inconsistent Naming

AgentToolsListCommand lives in Commands/Skills/ but is registered under the agents/tools command group. Per the {Resource}{SubResource}{Operation}Command naming pattern it should be AgentsToolsListCommand, placed in Commands/Agents/ alongside AgentsToolsGetCommand and AgentsToolsCreateCommand.

[g2vinay]

BLK-06 · SkillsDeleteCommand Calls DeleteAgentToolAsync — Functional Bug

SkillsDeleteCommand calls _sreAgentService.DeleteAgentToolAsync(endpoint, options.Name!, ...). This deletes a tool, not a skill. There is no DeleteSkillAsync method anywhere in ISreAgentService or SreAgentService.

The SRE Agent API has separate endpoints:

• /api/v2/extendedAgent/tools/{name} DELETE — tools
• /api/v2/extendedAgent/skills/{name} DELETE — skills

This looks like a functional bug: deleting a skill silently deletes a tool instead. DeleteSkillAsync must be added to ISreAgentService and SreAgentService, and SkillsDeleteCommand must call it.
Double check and implement.

[g2vinay]

BLK-07 · Unit Tests Are Shallow — Mandatory Patterns Missing for 53+ Commands

Per new-command.md, the required test patterns per command are:

• ExecuteAsync_ValidatesInputCorrectly — Theory with valid and invalid arg combinations
• ExecuteAsync_DeserializationValidation
• ExecuteAsync_HandlesServiceErrors — mocked service exception → correct HTTP status
• BindOptions_BindsOptionsCorrectly — verifies specific option values are correctly bound

Only AgentsListCommandTests implements this. All other 53 test files contain only:

void Constructor_InitializesCommandCorrectly() { Assert.NotNull(command.Name); }
void RegisterOptions_AddsExpectedOptions() { Assert.NotNull(command.Options); }

Assert.NotNull(command.Options) always passes regardless of whether the correct options are registered — the base class always provides a non-null collection. These stubs provide no meaningful coverage.

---

[BandaruDheeraj]

@g2vinay — addressed all blocking and high-severity findings from your review in 9f6d2193 except the three live-test items (BLK-01/02/03), which are deferred until I get access to a tenant where I can record the live tests. Build is clean (0 errors, 0 warnings) and 449/449 unit tests pass.

Blocking (non-live-test)

ID	Concern	Fix
BLK-04	Multiple Options classes per file	Split CommonPromptOptions.cs, ScheduledTasksOptions.cs, IncidentOptions.cs, ThreadsOptions.cs, DocsOptions.cs, WorkflowOptions.cs into one class per file (~30 new files under src/Options/<group>/).
BLK-05	azmcp-commands.md / e2eTestPrompts.md not updated	Added entries for all new SreAgent commands to servers/Azure.Mcp.Server/docs/azmcp-commands.md (+229 lines) and e2eTestPrompts.md (+38 lines), preserving alphabetical order.
BLK-06	SkillsDeleteCommand calls DeleteAgentToolAsync — deletes a tool, not a skill	Added DeleteSkillAsync to ISreAgentService and SreAgentService (DELETE /api/v2/extendedAgent/skills/{name}). SkillsDeleteCommand now calls it.
BLK-07	Unit tests are shallow stubs	Beefed up tests across the ported commands: ExecuteAsync_ValidatesInputCorrectly (Theory with valid/invalid args), ExecuteAsync_DeserializationValidation, ExecuteAsync_HandlesServiceErrors (mocked service exception → expected status), BindOptions_BindsOptionsCorrectly (verifies the typed option binding via SreAgentOptionDefinitions.X.Name). 449 tests passing.
BLK-08	CODEOWNERS not updated	Added tools/Azure.Mcp.Tools.SreAgent/ entry under # PRLabel: %tools-SreAgent.

High severity

ID	Concern	Fix
HIGH-01	SreAgentSetup not sealed	Changed to public sealed class SreAgentSetup : IAreaSetup.
HIGH-02	Ported delete commands return HTTP 200 with error text via SetTextResult	Switched the user-error paths to throw new InvalidOperationException("Refusing to delete ...: destructive operation requires --confirm true.") — matches the existing AgentsDeleteCommand/ConnectorsDeleteCommand pattern and now correctly maps to HTTP 4xx via HandleException. Success-path SetTextResult calls retained.
HIGH-04	Duplicate option definitions across SreAgentOptionDefinitions.cs and SreAgentPortedOptionDefinitions.cs	Deleted SreAgentPortedOptionDefinitions.cs; consolidated all definitions into SreAgentOptionDefinitions.cs. All call sites now bind via the typed OptionDefinitions.Foo.Name pattern.
HIGH-06	SSRF — agentEndpoint from ARM not validated against *.azuresre.ai	Added ValidateDataPlaneEndpoint(Uri) in SreAgentService.cs; rejects any endpoint whose host does not end with .azuresre.ai (case-insensitive). Called before GetTokenAsync in CallDataPlaneAsync and again in UploadMemoryAsync.
HIGH-07	PauseScheduledTaskAsync / ResumeScheduledTaskAsync pass null body to POST	Both now pass "{}" as jsonBody for consistency with ActivateThreadHookAsync and other action-POST data-plane calls.
HIGH-08	AgentToolsListCommand in Commands/Skills/ with inconsistent name	Renamed to AgentsToolsListCommand and moved to Commands/Agents/ alongside AgentsToolsGetCommand / AgentsToolsCreateCommand. Old file deleted; test file moved to match.

Deferred

• BLK-01, BLK-02, BLK-03 (LiveTests project, test-resources.bicep, test-resources-post.ps1, assets.json) — staying out of this commit until I have access to a tenant where I can deploy a Microsoft.App/agents resource and record live tests. Will address in a follow-up PR or commit on this branch once I get access.

Ready for re-review. 🙏