Paul

This commit delivers a **bug fix** for the **Copilot extension's node tools**, specifically refining the logic for directory confirmation. The `isDirExternalAndNeedsConfirmation` function in `toolUtils.ts` is updated to correctly identify subdirectories within **skill folders** as internal, preventing unnecessary external access confirmation prompts. New test cases have been added to `toolUtils.spec.ts` to validate this behavior, ensuring a smoother user experience when interacting with skill folder contents. This resolves an issue where internal skill folder paths were incorrectly flagged as requiring user confirmation.

This commit performs a significant **refactoring** and **cleanup** of the **Copilot extension's slash command processing and telemetry**. It introduces a new utility, `parseSlashCommand`, in `chatVariablesCollection.ts` to consistently extract slash command IDs from user queries and prompt files. This utility is then integrated into `chatParticipantTelemetry.ts` to streamline telemetry collection, removing a dependency on `ICustomInstructionsService`, and into `chatVariables.tsx` to eliminate duplicated parsing logic. The changes improve the consistency and robustness of **slash command handling** and result in cleaner telemetry within the **Copilot chat experience**.

This commit introduces a **new capability** by enhancing the `_listExtensionPromptFiles` API command within the **Extension Host API**. It now includes the `extensionId` as part of the `IExtensionPromptFileResult`, allowing consumers to identify the originating extension for each prompt file. This change primarily affects the **Chat prompt syntax contribution**, ensuring that prompt files listed via this API provide crucial context about their source. The enhancement enables more robust management and filtering of prompt files based on their owning extension.

This commit introduces a **new capability** to support **dynamic prompt variables** within the **Copilot chat system**. It implements a `PromptVariablesService` to resolve and inject context-specific information, such as `VSCODE_USER_PROMPTS_FOLDER` or `VSCODE_TARGET_SESSION_LOG`, directly into **system messages** and user prompts. This **enhances the flexibility and context-awareness** of **Copilot CLI sessions** and **agent prompts**, centralizing variable resolution and improving the overall prompt generation mechanism. Additionally, it includes **refactoring** to remove direct placeholder replacements and updates **documentation** for the new variable names.

This commit **enhances developer experience** by enabling **Intellisense support** for the `troubleshoot skill` module. It provides code completion, parameter information, and other helpful suggestions when interacting with this specific skill's configuration or programmatic interface. This **new capability** streamlines development and usage of the troubleshooting features, making it easier to write and debug related code or commands. The change primarily impacts developers and advanced users interacting with the `troubleshoot skill`'s definition or invocation.

This commit introduces a **new capability** for detailed performance monitoring within the **Copilot chat extension** by adding numerous performance markers. It establishes a dedicated utility in `src/util/common/performance.ts` for managing both session-scoped and global performance marks, complete with unit tests. Markers are strategically placed to measure critical operations such as **extension activation**, **chat participant handling**, and the duration of **prompt building and LLM fetching** within the tool calling loop. This enhancement provides crucial observability for diagnosing and optimizing the responsiveness and efficiency of the chat experience.

This commit introduces comprehensive **performance marking** for the **Chat feature**, enabling detailed instrumentation of its lifecycle. It adds new utilities in `src/vs/workbench/contrib/chat/common/chatPerf.ts` to mark key events such as request initiation, agent invocation, first token reception, and completion, alongside a new `clearMarks` function in `src/vs/base/common/performance.ts`. These marks are integrated across various chat components, including the widget, service, model, and agent participants, to measure critical timings like LLM round-trip and UI rendering. This **new capability** provides crucial data for identifying performance bottlenecks and optimizing the user experience within the chat interface.

This commit introduces a **new capability** to the **Copilot chat extension's hook service**, enabling more permissive handling of hook event names. It implements **one-way compatibility** for specific event types, such as recognizing 'Stop' as `SubagentStop` and 'SessionStart' as `SubagentStart`. This is achieved by integrating a new `isCompatibleHookEventName` function and `compatibleHookEventNames` map into the `executeHook` method of `ChatHookService`. This **enhancement** improves the flexibility and backward compatibility of the **chat hook system**, ensuring broader recognition of event names without breaking existing implementations. New test cases in `chatHookService.spec.ts` validate this compatibility logic.

This commit **enhances the chat feature's skill discovery mechanism** by prioritizing the **skill folder name** as the primary identifier. The `promptsServiceImpl.ts` file was updated to modify the `computeSkillDiscoveryInfo` logic, ensuring that when a skill's truncated name differs from its folder name, the folder name is used for identification. This **`grow` enhancement** improves the robustness of **chat skill resolution**, preventing issues where skills might not be correctly recognized due to naming inconsistencies. Updated test cases in `computeAutomaticInstructions.test.ts` and `promptsService.test.ts` validate this new fallback behavior.

This commit **fixes a potential runtime error** within the **Chat subagent functionality** by enhancing null safety. Specifically, it introduces a null check for the `arr` variable within the `some` method in `src/vs/workbench/contrib/chat/common/tools/builtinTools/runSubagentTool.ts` when evaluating subagent hooks. This **bug fix** prevents unexpected crashes or errors that could arise if `arr` were null during the hook checking process. The change improves the **robustness and stability of chat interactions** involving subagents, ensuring smoother operation.

This commit delivers a **bug fix** for the **Copilot extension's intent handling** by correcting how `Stop` and `SubagentStop` hooks process their output. It **refactors** the `StopHookOutput` and `SubagentStopHookOutput` interfaces in `chatHookService.ts` to consistently nest specific data under a new `hookSpecificOutput` field. Consequently, the parsing logic within the `toolCallingLoop.ts` was updated to correctly access decision and reason information from this new nested structure, ensuring the agent properly interprets these critical signals. New tests were added to validate the correct behavior of these hooks, improving the reliability of the **tool calling process**.

This commit **enhances the Chat prompt syntax service** by providing more comprehensive information during the skill discovery process. It modifies the `promptsServiceImpl.ts` file to include **source folder diagnostics** when skills are being identified, specifically by calling `_collectSourceFolderDiagnostics`. This **new capability** ensures that the chat feature has a richer context about available skills, which can lead to more accurate and relevant prompt suggestions or responses.

This commit performs a **maintenance refactoring** to standardize progress indication within the **Chat Debug Logs View** of the **Agent Debug Panel**. It replaces a custom-implemented shimmer effect with the platform's built-in `ProgressBar` component, specifically impacting `chatDebugLogsView.ts` and `chatDebug.css`. This change improves UI consistency and reduces custom code by leveraging existing components, leading to a more unified user experience for progress feedback in the debug panel.

This commit introduces **new telemetry events** across several **Copilot extensions** to enhance observability of debug panel scenarios and chat interactions. It adds detailed metrics for **chat debug file logging operations**, performance data for the **OTel debug log provider**'s compaction and span conversion, and telemetry for **chat slash commands**, hashing user-defined commands for privacy. This **new capability** provides crucial insights into how users interact with the **Copilot chat debug panel** and its features, aiding in future improvements and issue diagnosis. The changes affect the `chatDebugFileLoggerService`, `chatParticipantTelemetry`, and `otelChatDebugLogProvider` modules.

This commit **fixes a bug** in the **Copilot extension's prompt handling** for slash commands, specifically addressing how prompt files are referenced. It **refactors** the `UserQuery` component within the `extensions/copilot/src/extension/prompts/node/panel/chatVariables.tsx` file to utilize new helper functions, `getPromptFileSlashCommandId` and `buildSlashCommandUserMessage`. These new helpers ensure the correct identification of prompt file slash command IDs and accurate construction of user messages based on these commands. This **refactoring** and **bug fix** improves the reliability and correctness of **slash command functionality** within the Copilot chat panel, with new unit tests added for the helper functions.

This commit **reverts** a previous change that introduced chat performance markers within the **Copilot extension**. Specifically, it removes performance instrumentation calls from the `baseActivate` function, which manages extension activation and initialization, and from the `render` and `getOrCreateGlobalAgentContext` functions within the agent prompt system. This **maintenance** action effectively discontinues the collection of specific performance metrics related to chat interaction rendering and context retrieval. The change simplifies the codebase by removing these temporary profiling hooks, impacting the **Copilot extension's core activation logic** and its **agent prompt system**.

This commit **reverts** a previous change that introduced performance markers within the **chat feature** of the workbench. It specifically removes instrumentation from key interaction points, including the `acceptInput` function in the `chatWidget`, `sendRequest`, `resendRequest`, and `_sendRequestAsync` in the `chatService`, and `invokeAgent` in `chatAgents`. This **maintenance** action undoes the addition of performance measurement logic, simplifying the codebase by removing the associated overhead. The immediate impact is the cessation of performance data collection for these specific chat operations, effectively rolling back the performance monitoring capability.

This commit introduces comprehensive **debugging capabilities** for **Copilot chat event logs** by integrating **OpenTelemetry tracing** into hook command executions. It enables the capture of detailed input, output, and results for executed hooks, which are then converted into debug events for a dedicated **chat debug panel**. This **new capability** primarily affects the **Copilot chat extension's hook execution service** and includes a minor **bug fix** for session ID resolution in `ChatParticipantRequestHandler`. The changes significantly improve the diagnosability of chat interactions by providing granular visibility into hook lifecycle events.

This commit introduces a **new capability** to the **Chat Debugging** feature by adding **hooks execution logging** to the **chat debug details panel**. It defines new types such as `ChatDebugHookResult` and `ChatDebugEventHookContent` across the extension host and main thread protocols, enabling extensions to provide detailed information about chat hook events. The `chatDebug` API proposal is updated, and the `chatDebugDetailPanel.ts` is enhanced to render this new content, offering users a more comprehensive understanding of chat interaction flows.

This commit introduces **performance markers** across critical components of the **Copilot chat** experience to enable detailed **performance monitoring**. This **maintenance** work adds tracking to key areas such as `chatParticipants` handling, `conversationFeature` activation, the main extension activation process, and the `toolCallingLoop` execution. Furthermore, markers are now present for `agentPrompt` creation and `customInstructions` preparation, providing essential data for identifying and optimizing bottlenecks within the chat system.