SDK 1.0.5 + perf fixes (reasoning effort, task sub-agent) + gpt-5.6-sol with Priority processing - #54
Merged
Conversation
Migrate from GitHub.Copilot.SDK 1.0.0-beta.4 to 1.0.5. API changes: namespace GitHub.Copilot.SDK -> GitHub.Copilot; CopilotClientOptions.CopilotHome -> BaseDirectory; OnPermissionRequest -> PermissionHandler.ApproveAll (replaces PermissionRequestResult); CopilotSession.GetMessagesAsync -> GetEventsAsync; SessionListFilter.Cwd -> WorkingDirectory; SessionMetadata.Context.Cwd -> WorkingDirectory; SessionConfig.Tools is now ICollection<AIFunctionDeclaration> so tool-list types widened from List<AIFunction> to List<AIFunctionDeclaration>. Behavior-preserving. Verified: dotnet build Release GREEN (0/0). Supersedes #51. [skip ci] to avoid an unattended test-slot deploy; CI runs on merge to main.
…agent, gpt-5.6-sol + Priority processing, dep bumps Slowness fixes from App Insights investigation (12.7-min chat turns): (1) ReasoningEffort now read from AzureOpenAI:ReasoningEffort, default 'high' — hardcoded 'xhigh' produced 8+ min single LLM round-trips; (2) Copilot CLI 'task' sub-agent excluded via SessionConfig.ExcludedTools — one call looped 8 min in prod; (3) default model gpt-5.6-sol 2026-07-09 with serviceTier Priority (new Bicep param, default Priority) for faster TTFT; (4) NuGet bumps: IdentityModel 8.19.1 x2 (#41 #42), OpenTelemetry 1.16.0 x2 (#44). Verified: dotnet build Release green; bicep build clean.
Ali Reza Farahnak (alfarahn)
changed the title
This was referenced Jul 11, 2026
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Three things in one shippable unit — SDK GA upgrade, the production slowness fixes from the App Insights investigation, and the model refresh to
gpt-5.6-solwith Priority processing. Also applies the pending NuGet Dependabot bumps (supersedes #41, #42, #44, #51).1. GitHub.Copilot.SDK
1.0.0-beta.4→1.0.5(GA)API surface confirmed by reflecting the shipped assembly:
GitHub.Copilot.SDKGitHub.CopilotCopilotClientOptions.CopilotHomeBaseDirectorynew PermissionRequestResult { Kind = Approved }PermissionHandler.ApproveAllGetMessagesAsync()GetEventsAsync()SessionListFilter.CwdSessionListFilter.WorkingDirectorySessionMetadata.Context.Cwd.Context.WorkingDirectoryList<AIFunction>List<AIFunctionDeclaration>2. Performance fixes (prod App Insights showed 12.7-minute chat turns)
AzureOpenAI:ReasoningEffort, defaulthigh(was hardcodedxhigh). Telemetry showed singlexhighLLM round-trips of 509 s and 489 s.xhighremains available as an explicit opt-in.tasksub-agent excluded (SessionConfig.ExcludedTools = ["task"]) — onetaskcall spawned a nested general-purpose agent that looped for 8.1 minutes in production. FinOps work uses direct tools only.3. Model refresh:
gpt-5.4→gpt-5.6-sol+ Priority processinggpt-5.6-sol(version2026-07-09, GlobalStandard).infra/modules/aoai.bicepnow setsproperties.serviceTier(newserviceTierparam, default'Priority') — the Foundry portal's "Priority processing" toggle, for faster time-to-first-token. Verified against theMicrosoft.CognitiveServices/accounts/deployments@2026-03-01ARM schema.gpt-5.6-sol(1M TPM, Priority, DefaultV2 guardrails) already created onfinops-agent-aivia ARM.4. Dependency updates
Microsoft.IdentityModel.JsonWebTokens+Protocols.OpenIdConnect8.18.0→8.19.1(supersedes Bump Microsoft.IdentityModel.JsonWebTokens from 8.18.0 to 8.19.1 #41, Bump Microsoft.IdentityModel.Protocols.OpenIdConnect from 8.18.0 to 8.19.1 #42)OpenTelemetry+OpenTelemetry.Api1.15.3→1.16.0(supersedes Bump OpenTelemetry and OpenTelemetry.Api #44)Type of change
Checklist
dotnet build) — Release, 0 warnings / 0 errorsnpm run buildinsrc/Dashboard/frontend) — no frontend changes in this PR.github/copilot-instructions.mdif behaviour or structure changedCHANGELOG.mdunder[Unreleased]bicep build)Related issues
Closes #51