Logan Ramos

This commit **refactors and enhances** the **account management** and **entitlement fetching** logic to reduce unnecessary requests to entitlement endpoints. It introduces a `chat_enabled` property to the `IEntitlementsData` interface and modifies the `IAccountPolicyData` to track `entitlementsFetchedAt`. The core change involves implementing **conditional token entitlement fetching**, ensuring that entitlements are only requested when necessary, particularly for chat features, by removing the `useExistingEntitlements` parameter. This **feature enhancement** optimizes the `defaultAccount` service by **reducing network traffic** and improving performance for users.

This commit delivers a **bug fix** to the **account services** within the workbench, specifically addressing the interpretation of the `mcp` flag. It **corrects the logic** in `src/vs/workbench/services/accounts/browser/defaultAccount.ts` so that the `mcp` flag is now only enabled when explicitly set to `'1'`. This change prevents the previous behavior where the flag was enabled for any value not equal to `'0'`, thereby resolving the issue of **over requesting of the MCP registry** and ensuring more accurate control over account-related operations.

This commit implements a **fix** to the **AI Edit Telemetry service** by **removing the use of `TelemetryTrustedValue` for `modelId`** and introducing explicit sanitization. It specifically modifies the `logEditSuggested` and `logEditAccepted` telemetry events within `src/vs/workbench/contrib/editTelemetry/browser/telemetry/aiEditTelemetry/aiEditTelemetryServiceImpl.ts` to replace slashes in the `modelId` string with underscores before transmission. This **improves data handling and privacy** by ensuring `modelId` values are consistently formatted and not inadvertently treated as fully trusted. The change prevents potential issues with telemetry data integrity or the exposure of sensitive path information, ensuring more secure and consistent reporting of `modelId` for AI-assisted editing features.

This commit **refactors** and **enhances** the **Copilot extension's authentication mechanism** by centralizing the determination of VS Code organization membership. Specifically, the `extensions/copilot/src/platform/authentication/common/copilotToken.ts` module now exports the `containsVSCodeOrg` function, making this critical logic reusable. The `extensions/copilot/src/platform/authentication/node/copilotTokenManager.ts` module was updated to directly leverage this exported function, ensuring a consistent and maintainable approach to checking **VS Code team membership** during the authentication process. This change improves the robustness of **Copilot token management** by standardizing how organizational affiliation is verified.

This commit **addresses comments and implements fixes** related to **chat entitlements**, particularly for the **EDU SKU**. It **updates the numeric values** of the `ChatEntitlement` enum members, including `EDU`, `Pro`, and `Enterprise`, within the **Chat Entitlement Service** (`chatEntitlementService.ts`). Concurrently, the **chat model picker action bar provider** in `modelPickerActionItem.ts` is enhanced to correctly recognize and check for the `ChatEntitlement.EDU` status. This **maintenance update** ensures accurate handling and display of chat models based on a user's specific entitlement level.

This commit introduces a **new capability** by integrating the `EDU` (Educational) entitlement into the **Chat feature's** entitlement handling system. It updates the `src/vs/workbench/services/chat/common/chatEntitlementService.ts` to recognize `EDU` as a distinct plan, affecting how user plans are identified and context keys are updated. Consequently, users with the `EDU` entitlement are now granted access to various premium chat functionalities, including the ability to **manage language models**, view the 'Manage Paid Premium Requests' button when quotas are exceeded, and generally participate in chat features previously restricted to other premium plans. This ensures the **Chat UI**, actions, and management components correctly reflect and enable features for `EDU` plan holders.

This commit **enhances the Copilot extension** by establishing 'medium' as the **default reasoning effort** for AI model interactions. Specifically, the `createResponsesRequestBody` function within `extensions/copilot/src/platform/endpoint/node/responsesApi.ts` will now automatically apply a 'medium' effort level if no explicit reasoning effort is specified in the request options. This **configuration change** ensures more consistent and predictable behavior for the reasoning process, preventing undefined effort levels and standardizing the default AI model interaction. The change improves robustness by guaranteeing a sensible default for all requests originating from this part of the system.

This commit introduces **telemetry** to the **Copilot extension** to enhance understanding of user behavior when encountering rate limits. It specifically tracks user actions such as switching to an auto-model or retrying requests within the `chatParticipants.ts` module, using new events in `handleRequest` and `handleRetry`. Additionally, a new helper function `isContinueOnError` is added to `specialRequestTypes.ts` to identify chat requests that are continuations after an error. This **new capability** provides crucial data for improving the auto rate limit swap feature and supports more robust error handling in chat interactions.

This commit introduces a **new capability** to the **Copilot automode service** by implementing `invalidateRouterCache` and ensuring the router re-evaluates its state after invalidation. It integrates this service into the **agent intent handling**, specifically triggering cache invalidation within `AgentIntent` when conversations are compacted or background tool call results are processed. This change also involves **refactoring** several intent invocations (`AskAgentIntentInvocation`, `EditCode2IntentInvocation`, `NotebookEditorIntentInvocation`) to properly inject the `automodeService` dependency. The update improves the responsiveness and accuracy of **Copilot's automatic mode selection** by preventing stale routing decisions after significant conversation state changes.

This commit introduces a **1-second timeout** for the **Copilot extension's router decision fetching**, addressing potential hanging requests and improving system resilience. The `fetchRouterDecision` function now utilizes an `AbortController` to prevent indefinite waits, ensuring that the **automode endpoint resolution** does not block. If the fetch times out, the system will **fall back to a default selection**, and telemetry in `automodeService.ts` will record a specific `routerTimeout` reason. New tests in `automodeService.spec.ts` confirm this expected timeout and fallback behavior, making the **automode service** more robust.

This commit introduces a **new capability** to the **Copilot extension's chat platform** by enhancing **rate limit messages** to include an estimated retry time. It **removes the `hideRateLimitTimeEstimate` option** from various internal functions and request handlers, such as `getAgentRequestOptions` and `getErrorDetailsFromChatFetchError`, as this option is no longer relevant. This change provides users with clearer guidance when encountering rate limits, improving the overall user experience by informing them when they can expect to retry their requests. The work involves both a **feature addition** and **refactoring** of the rate limit handling logic within the `extensions/copilot` module.

This commit **enhances the Claude chat integration** by ensuring that the available Claude language models respect the `showInModelPicker` flag, preventing internal or non-selectable models from appearing in the user interface. It introduces a new `getEndpoints` method within the **Claude code models service** (`claudeCodeModels.ts`) to filter chat endpoints based on this flag. The `ClaudeLanguageModelServer` is **refactored** to utilize this new service, improving dependency management and providing a more curated selection of models. This **feature enhancement** ensures users see an accurate and relevant list of Claude models in the **model picker UI**.

This commit delivers a **bug fix** for the **Copilot automode endpoint resolution service**, preventing the auto router from repeatedly retrying failed requests after an initial failure. It **refactors** the `automodeService.ts` file by enhancing the cache to track router fallback reasons, ensuring the system intelligently handles transient failures without redundant re-evaluations. This improvement to the **endpoint routing logic** directly impacts the stability and efficiency of Copilot's backend communication by avoiding unnecessary resource consumption. Accompanying changes in `automodeService.spec.ts` include new test cases to verify the correct router re-evaluation behavior.

This commit implements a **bug fix** within the **Copilot extension's telemetry service** to prevent the unintended redaction of callsite information for `fetch` telemetry events. By modifying `extensions/copilot/src/platform/telemetry/vscode-node/telemetryServiceImpl.ts`, the call site is now explicitly marked as a trusted value, ensuring its inclusion in telemetry reports. This change addresses an issue where valuable context about the origin of `fetch` operations was being removed. Ultimately, it improves the accuracy and completeness of collected **telemetry data**, aiding in better analysis and debugging of the extension's network interactions.

This commit implements a **maintenance fix** to resolve a **GDPR classification conflict** within the **TypeScript language features extension**. It specifically updates the GDPR classification comments for telemetry events `installSuccess` and `typingsInstallerVersion` in `typescriptServiceClient.ts`. The classification is changed from `PerformanceAndHealth` to `FeatureInsight`, ensuring more accurate **GDPR compliance** for these specific telemetry data points.

This commit **refines the user interface text** within the **Copilot extension's prompt handling** by updating the labels for the 'Switch to Auto' confirmation buttons. Specifically, the text in `extensions/copilot/src/extension/prompt/node/defaultIntentRequestHandler.ts` has been made more concise. This is a **usability improvement** aimed at enhancing clarity and conciseness for users. The change provides a more streamlined and direct user experience when interacting with the Copilot's intent recognition features that prompt for switching to auto mode.

This commit introduces a **new capability** to the **Copilot extension**, enabling it to automatically handle **model rate limits** during chat requests. When a rate limit is encountered, the system will now **automatically switch to the 'Auto' model and retry the request**, significantly improving the resilience and continuity of user interactions. This behavior is configurable via a new setting, `RateLimitAutoSwitchToAuto`, and involves updates across the **conversation, prompt, and configuration services** to manage the automatic model switching and retry logic.

This commit **refines the model selection experience** within the **Copilot extension's Claude chat sessions**. It introduces a **filter to hide Claude chat endpoints** that are not designated as "model picker eligible" in `claudeLanguageModelServer.ts`. This **enhancement** ensures that users are presented with a curated and relevant list of Claude models, specifically those that support the Messages API, thereby improving usability and reducing clutter in the model picker. The change primarily impacts the visibility and selection of available Claude models for chat interactions.

This commit **enhances the user experience for rate limiting** within the **Copilot extension's chat platform**. It introduces **new rate limit error codes** and significantly **improves the specificity of user-facing messages** generated by the `getRateLimitMessage` function. Users will now receive more informative feedback, including helpful 'Learn More' links, when encountering rate limits, thereby improving guidance and reducing confusion. This is a **feature enhancement** focused on better error communication in the **chat subsystem**.

This commit **optimizes telemetry behavior** by preventing the **`telemetryService`** from sending **`fetcher telemetry`** when the product is running in a **stable quality** environment. It modifies the telemetry sending logic within both the browser and Electron-browser implementations of the **workbench telemetry service**. This **maintenance** change reduces unnecessary data transmission and enhances user privacy for users on stable builds.