[pull] main from microsoft:main

extensions/copilot/package.json

@@ -2684,6 +2684,12 @@
 				"category": "Chat",
 				"enablement": "github.copilot.sessionSearch.enabled && config.chat.sessionSync.enabled"
 			},
+			{
+				"command": "github.copilot.chronicle.reindex",
+				"title": "%github.copilot.command.chronicle.reindex%",
+				"category": "Chat",
+				"enablement": "github.copilot.sessionSearch.enabled"
+			},
 			{
 				"command": "github.copilot.nes.captureExpected.start",
 				"title": "Record Expected Edit (NES)",
@@ -4499,16 +4505,6 @@
 							"experimental"
 						]
 					},
-					"github.copilot.chat.agentHistorySummarizationInline": {
-						"type": "boolean",
-						"default": true,
-						"markdownDescription": "%github.copilot.config.agentHistorySummarizationInline%",
-						"tags": [
-							"advanced",
-							"experimental",
-							"onExp"
-						]
-					},
 					"github.copilot.chat.useResponsesApiTruncation": {
 						"type": "boolean",
 						"default": false,
@@ -4678,6 +4674,14 @@
 							"advanced"
 						]
 					},
+					"github.copilot.chat.agent.modelDetails.enabled": {
+						"type": "boolean",
+						"default": true,
+						"markdownDescription": "%github.copilot.config.chat.agent.modelDetails.enabled%",
+						"tags": [
+							"advanced"
+						]
+					},
 					"github.copilot.chat.cli.planCommand.enabled": {
 						"type": "boolean",
 						"default": true,

extensions/copilot/package.nls.json

@@ -172,6 +172,7 @@
 	"copilot.chronicle.standup.description": "Generate a standup report from recent chat sessions",
 	"copilot.chronicle.tips.description": "Get personalized tips based on your chat session usage patterns",
 	"github.copilot.command.sessionSync.deleteSessions": "Delete Session Sync Data",
+	"github.copilot.command.chronicle.reindex": "Reindex Sessions",
 	"copilot.chronicle.reindex.description": "Rebuild the local session index from stored session logs. Add 'force' to re-process already indexed sessions.",
 	"github.copilot.config.sessionSearch.enabled": "Enable session search and /chronicle commands. This is a team-internal setting.",
 	"github.copilot.config.sessionSearch.localIndex.enabled": "Enable local session tracking. When enabled, Copilot tracks session data locally for /chronicle commands.",
@@ -398,7 +399,6 @@
 	"github.copilot.config.instantApply.shortContextLimit": "Token limit for short context instant apply.",
 	"github.copilot.config.summarizeAgentConversationHistoryThreshold": "Threshold for compacting agent conversation history.",
 	"github.copilot.config.agentHistorySummarizationMode": "Mode for agent history summarization.",
-	"github.copilot.config.agentHistorySummarizationInline": "Summarize conversation inline within the agent loop instead of a separate LLM call, maximizing prompt cache hits.",
 	"github.copilot.config.useResponsesApiTruncation": "Use Responses API for truncation.",
 	"github.copilot.config.enableReadFileV2": "Enable version 2 of the read file tool.",
 	"github.copilot.config.enableAskAgent": "Enable the Ask agent for answering questions.",
@@ -413,6 +413,7 @@
 	"github.copilot.config.cli.showExternalSessions": "Show sessions created by other applications.",
 	"github.copilot.config.cli.planExitMode.enabled": "Enable Plan Mode exit handling in Copilot CLI.",
 	"github.copilot.config.cli.autoModel.enabled": "Enable the Auto model option in Copilot CLI, which automatically selects the best model for each request. Requires VS Code reload.",
+	"github.copilot.config.chat.agent.modelDetails.enabled": "Show model details (model name and request multiplier) on agent chat responses when using Copilot CLI or Claude agent in VS Code. Requires VS Code reload to update already loaded sessions.",
 	"github.copilot.config.cli.planCommand.enabled": "Enable the /plan command in Copilot CLI to create implementation plans before coding.",
 	"github.copilot.config.cli.lazyLoadSessionItem.enabled": "Enable lazy loading of session items in Copilot CLI. Requires VS Code reload.",
 	"github.copilot.config.cli.aiGenerateBranchNames.enabled": "Enable AI-generated branch names in Copilot CLI.",

extensions/copilot/src/extension/chatSessions/claude/node/claudeCodeModels.ts

@@ -97,6 +97,9 @@ export class ClaudeCodeModels extends Disposable implements IClaudeCodeModels {
 				maxInputTokens: endpoint.modelMaxPromptTokens,
 				maxOutputTokens: endpoint.maxOutputTokens,
 				pricing: multiplier ?? (endpoint.tokenPricing ? formatPricingLabel(endpoint.tokenPricing) : undefined),
+				inputCost: endpoint.tokenPricing?.inputPrice,
+				outputCost: endpoint.tokenPricing?.outputPrice,
+				cacheCost: endpoint.tokenPricing?.cacheReadTokenPrice,
 				multiplierNumeric: endpoint.multiplier,
 				tooltip,
 				isUserSelectable: true,

extensions/copilot/src/extension/chatSessions/common/chatSessionMetadataStore.ts

@@ -56,6 +56,14 @@ export interface RequestDetails {
 	/** Mode instructions for this request (excluding toolReferences). */
 	modeInstructions?: StoredModeInstructions;
 
+	/**
+	 * The concrete model id that produced the response for this request, as reported by the
+	 * SDK's `assistant.usage` event. Captured so that on session reload we can render the
+	 * correct model details (e.g. for `auto`, where the resolved model is not otherwise
+	 * recoverable from the persisted SDK event log).
+	 */
+	responseModelId?: string;
+
 	/** Checkpoint reference for this request (primary workspace). */
 	checkpointRef?: string;
 

extensions/copilot/src/extension/chatSessions/copilotcli/common/copilotCLITools.ts

@@ -518,22 +518,57 @@ export interface RequestIdDetails {
 	readonly requestId: string;
 	readonly toolIdEditMap: Record<string, string>;
 	readonly modeInstructions?: StoredModeInstructions;
+	readonly responseModelId?: string;
 }
 
 /**
  * Build chat history from SDK events for VS Code chat session
  * Converts SDKEvents into ChatRequestTurn2 and ChatResponseTurn2 objects
  */
-export function buildChatHistoryFromEvents(sessionId: string, modelId: string | undefined, events: readonly SessionEvent[], getVSCodeRequestId: (sdkRequestId: string) => RequestIdDetails | undefined, delegationSummaryService: IChatDelegationSummaryService, logger: ILogger, workingDirectory?: URI, defaultModeInstructionsForLastRequest?: StoredModeInstructions, lastResponseDetails?: string): (ChatRequestTurn2 | ChatResponseTurn2)[] {
+export function buildChatHistoryFromEvents(sessionId: string, modelId: string | undefined, events: readonly SessionEvent[], getVSCodeRequestId: (sdkRequestId: string) => RequestIdDetails | undefined, delegationSummaryService: IChatDelegationSummaryService, logger: ILogger, workingDirectory?: URI, defaultModeInstructionsForLastRequest?: StoredModeInstructions, modelDetailsById?: ReadonlyMap<string, string>): (ChatRequestTurn2 | ChatResponseTurn2)[] {
 	const turns: (ChatRequestTurn2 | ChatResponseTurn2)[] = [];
 	let currentResponseParts: ExtendedChatResponsePart[] = [];
 	const pendingToolInvocations = new Map<string, [ChatToolInvocationPart | ChatResponseMarkdownPart | ChatResponseThinkingProgressPart, toolData: ToolCall, parentToolCallId: string | undefined]>();
 
 	let details: RequestIdDetails | undefined;
 	let isFirstUserMessage = true;
+	let currentModelId = modelId;
+	let currentResponseModelId: string | undefined;
+	let currentRequestTurnIndex: number | undefined;
 	const currentAssistantMessage: { chunks: string[] } = { chunks: [] };
 	const processedMessages = new Set<string>();
 
+	function getModelDetails(modelId: string | undefined): string | undefined {
+		if (!modelId || !modelDetailsById) {
+			return undefined;
+		}
+		return modelDetailsById.get(modelId.trim().toLowerCase());
+	}
+
+	function createResultForModel(modelId: string | undefined) {
+		const details = getModelDetails(modelId);
+		return details ? { details } : {};
+	}
+
+	function flushResponseParts() {
+		if (currentResponseParts.length > 0) {
+			turns.push(new ChatResponseTurn2(currentResponseParts, createResultForModel(currentResponseModelId ?? currentModelId), ''));
+			currentResponseParts = [];
+		}
+		currentResponseModelId = undefined;
+		currentRequestTurnIndex = undefined;
+	}
+
+	function updateCurrentRequestModelId(modelId: string | undefined) {
+		if (currentRequestTurnIndex === undefined || !modelId) {
+			return;
+		}
+		const turn = turns[currentRequestTurnIndex];
+		if (turn instanceof ChatRequestTurn2 && turn.modelId !== modelId) {
+			turns[currentRequestTurnIndex] = new ChatRequestTurn2(turn.prompt, turn.command, turn.references, turn.participant, [...turn.toolReferences], turn.editedFileEvents, turn.id, modelId, turn.modeInstructions2);
+		}
+	}
+
 	function processAssistantMessage(content: string) {
 		// Extract PR metadata if present
 		const { cleanedContent, prPart } = extractPRMetadata(content);
@@ -562,16 +597,33 @@ export function buildChatHistoryFromEvents(sessionId: string, modelId: string |
 		}
 
 		switch (event.type) {
+			case 'session.start':
+			case 'session.resume': {
+				currentModelId = event.data.selectedModel ?? currentModelId;
+				break;
+			}
+			case 'session.model_change': {
+				currentModelId = event.data.newModel;
+				if (currentRequestTurnIndex !== undefined && currentResponseParts.length === 0) {
+					currentResponseModelId = currentModelId;
+					updateCurrentRequestModelId(currentModelId);
+				}
+				break;
+			}
+			case 'assistant.usage': {
+				currentModelId = event.data.model ?? currentModelId;
+				if (currentRequestTurnIndex !== undefined) {
+					currentResponseModelId = currentModelId;
+					updateCurrentRequestModelId(currentModelId);
+				}
+				break;
+			}
 			case 'user.message': {
 				if (isSyntheticUserMessage(event)) {
 					continue;
 				}
 				details = getVSCodeRequestId(event.id);
-				// Flush any pending response parts before adding user message
-				if (currentResponseParts.length > 0) {
-					turns.push(new ChatResponseTurn2(currentResponseParts, {}, ''));
-					currentResponseParts = [];
-				}
+				flushResponseParts();
 				// Filter out vscode instruction files from references when building session history
 				// TODO@rebornix filter instructions should be rendered as "references" in chat response like normal chat.
 				const references: ChatPromptReference[] = [];
@@ -675,7 +727,13 @@ export function buildChatHistoryFromEvents(sessionId: string, modelId: string |
 					}
 				}
 
-				turns.push(new ChatRequestTurn2(`${commandPrefix}${prompt}`, undefined, references, '', [], undefined, details?.requestId ?? event.id, modelId, modeInstructions2));
+				// Prefer the persisted resolved model id (from `assistant.usage`) so that on reload
+				// `auto` sessions show the actual model used to produce the response. Falls back to
+				// the currently tracked model id (from `session.start`/`session.model_change`).
+				const resolvedRequestModelId = details?.responseModelId ?? currentModelId;
+				currentResponseModelId = resolvedRequestModelId;
+				turns.push(new ChatRequestTurn2(`${commandPrefix}${prompt}`, undefined, references, '', [], undefined, details?.requestId ?? event.id, resolvedRequestModelId, modeInstructions2));
+				currentRequestTurnIndex = turns.length - 1;
 				break;
 			}
 			case 'assistant.message_delta': {
@@ -745,10 +803,7 @@ export function buildChatHistoryFromEvents(sessionId: string, modelId: string |
 	}
 
 	flushPendingAssistantMessage();
-
-	if (currentResponseParts.length > 0) {
-		turns.push(new ChatResponseTurn2(currentResponseParts, lastResponseDetails ? { details: lastResponseDetails } : {}, ''));
-	}
+	flushResponseParts();
 
 	return turns;
 }

extensions/copilot/src/extension/chatSessions/copilotcli/common/test/copilotCLITools.spec.ts

@@ -163,12 +163,92 @@ describe('CopilotCLITools', () => {
 				{ type: 'user.message', data: { content: 'Hello', attachments: [] } },
 				{ type: 'assistant.message', data: { content: 'Hi there' } }
 			];
-			const turns = buildChatHistoryFromEvents('', 'base', events, getVSCodeRequestId, delegationSummary, logger, undefined, undefined, 'Base • 2x');
+			const turns = buildChatHistoryFromEvents('', 'base', events, getVSCodeRequestId, delegationSummary, logger, undefined, undefined, new Map([['base', 'Base • 2x']]));
 			expect(turns).toHaveLength(2);
 			const responseTurn = turns[1] as ChatResponseTurn2;
 			expect(responseTurn.result).toEqual({ details: 'Base • 2x' });
 		});
 
+		it('uses session model changes for each rebuilt response turn', () => {
+			const modelDetails = new Map([
+				['opus-4.6', 'Opus 4.6 • 4x'],
+				['opus-4.7', 'Opus 4.7 • 4x'],
+				['gpt-5.4', 'GPT 5.4 • 2x'],
+				['gpt-5.3', 'GPT 5.3 • 1x'],
+			]);
+			const events: any[] = [
+				{ type: 'session.start', data: { selectedModel: 'opus-4.6' } },
+				{ type: 'user.message', id: 'u1', data: { content: 'First', attachments: [] } },
+				{ type: 'assistant.message', data: { content: 'One' } },
+				{ type: 'session.model_change', data: { previousModel: 'opus-4.6', newModel: 'opus-4.7' } },
+				{ type: 'user.message', id: 'u2', data: { content: 'Second', attachments: [] } },
+				{ type: 'assistant.message', data: { content: 'Two' } },
+				{ type: 'session.model_change', data: { previousModel: 'opus-4.7', newModel: 'gpt-5.4' } },
+				{ type: 'user.message', id: 'u3', data: { content: 'Third', attachments: [] } },
+				{ type: 'assistant.message', data: { content: 'Three' } },
+				{ type: 'session.model_change', data: { previousModel: 'gpt-5.4', newModel: 'gpt-5.3' } },
+				{ type: 'user.message', id: 'u4', data: { content: 'Fourth', attachments: [] } },
+				{ type: 'assistant.message', data: { content: 'Four' } },
+			];
+
+			const turns = buildChatHistoryFromEvents('', undefined, events, getVSCodeRequestId, delegationSummary, logger, undefined, undefined, modelDetails);
+
+			expect(turns.filter(turn => turn instanceof ChatRequestTurn2).map(turn => (turn as ChatRequestTurn2).modelId)).toEqual(['opus-4.6', 'opus-4.7', 'gpt-5.4', 'gpt-5.3']);
+			expect(turns.filter(turn => turn instanceof ChatResponseTurn2).map(turn => (turn as ChatResponseTurn2).result)).toEqual([
+				{ details: 'Opus 4.6 • 4x' },
+				{ details: 'Opus 4.7 • 4x' },
+				{ details: 'GPT 5.4 • 2x' },
+				{ details: 'GPT 5.3 • 1x' },
+			]);
+		});
+
+		it('uses assistant usage model for the active rebuilt response turn', () => {
+			const events: any[] = [
+				{ type: 'user.message', id: 'u1', data: { content: 'Hello', attachments: [] } },
+				{ type: 'assistant.message', data: { content: 'Hi' } },
+				{ type: 'assistant.usage', data: { model: 'gpt-5.4', inputTokens: 10, outputTokens: 5 } },
+			];
+
+			const turns = buildChatHistoryFromEvents('', undefined, events, getVSCodeRequestId, delegationSummary, logger, undefined, undefined, new Map([['gpt-5.4', 'GPT 5.4 • 2x']]));
+
+			expect(turns).toHaveLength(2);
+			expect((turns[0] as ChatRequestTurn2).modelId).toBe('gpt-5.4');
+			expect((turns[1] as ChatResponseTurn2).result).toEqual({ details: 'GPT 5.4 • 2x' });
+		});
+
+		it('uses persisted responseModelId to recover model details on reload for auto sessions', () => {
+			// Simulates a reloaded `auto` session: the SDK only persists `selectedModel: "auto"`
+			// (the `assistant.usage` event that carried the resolved model id is ephemeral and
+			// dropped from the persisted event log). The resolved model id was previously
+			// captured by the participant and stored via the chat session metadata store as
+			// `RequestDetails.responseModelId`, then surfaced through the `getVSCodeRequestId`
+			// callback. The reload path must use it to render the model footer details.
+			const events: any[] = [
+				{ type: 'session.start', data: { selectedModel: 'auto' } },
+				{ type: 'user.message', id: 'u1', data: { content: 'First', attachments: [] } },
+				{ type: 'assistant.message', data: { content: 'One' } },
+				{ type: 'user.message', id: 'u2', data: { content: 'Second', attachments: [] } },
+				{ type: 'assistant.message', data: { content: 'Two' } },
+			];
+			const detailsByEventId: Record<string, RequestIdDetails> = {
+				u1: { requestId: 'r1', toolIdEditMap: {}, responseModelId: 'gpt-5.4' },
+				u2: { requestId: 'r2', toolIdEditMap: {}, responseModelId: 'claude-opus-4.7' },
+			};
+			const lookup = (sdkRequestId: string) => detailsByEventId[sdkRequestId];
+
+			const turns = buildChatHistoryFromEvents('', 'auto', events, lookup, delegationSummary, logger, undefined, undefined, new Map([
+				['gpt-5.4', 'GPT 5.4 • 2x'],
+				['claude-opus-4.7', 'Claude Opus 4.7 • 4x'],
+			]));
+
+			expect(turns).toHaveLength(4);
+			expect(turns.filter(turn => turn instanceof ChatRequestTurn2).map(turn => (turn as ChatRequestTurn2).modelId)).toEqual(['gpt-5.4', 'claude-opus-4.7']);
+			expect(turns.filter(turn => turn instanceof ChatResponseTurn2).map(turn => (turn as ChatResponseTurn2).result)).toEqual([
+				{ details: 'GPT 5.4 • 2x' },
+				{ details: 'Claude Opus 4.7 • 4x' },
+			]);
+		});
+
 		it('converts file attachments to references on user messages', () => {
 			const events: any[] = [
 				{

extensions/copilot/src/extension/chatSessions/copilotcli/node/copilotCli.ts

@@ -66,6 +66,11 @@ export function formatModelDetails(model: CopilotCLIModelInfo): string {
 	return `${model.name}${model.multiplier ? ` • ${model.multiplier}x` : ''}`;
 }
 
+export function matchesCopilotCLIModel(model: Pick<CopilotCLIModelInfo, 'id' | 'name'>, modelId: string): boolean {
+	const normalizedModelId = modelId.trim().toLowerCase();
+	return model.id.trim().toLowerCase() === normalizedModelId || model.name.trim().toLowerCase() === normalizedModelId;
+}
+
 export const ICopilotCLISDK = createServiceIdentifier<ICopilotCLISDK>('ICopilotCLISDK');
 
 export const ICopilotCLIModels = createServiceIdentifier<ICopilotCLIModels>('ICopilotCLIModels');
@@ -114,7 +119,7 @@ export class CopilotCLIModels extends Disposable implements ICopilotCLIModels {
 		}
 		const models = await this.getModels();
 		modelId = modelId.trim().toLowerCase();
-		return models.find(m => m.id.toLowerCase() === modelId || m.name.toLowerCase() === modelId)?.id;
+		return models.find(m => matchesCopilotCLIModel(m, modelId))?.id;
 	}
 	public async getDefaultModel() {
 		// First item in the list is always the default model (SDK sends the list ordered based on default preference)
@@ -123,9 +128,9 @@ export class CopilotCLIModels extends Disposable implements ICopilotCLIModels {
 			return;
 		}
 		const defaultModel = models[0];
-		const preferredModelId = this.extensionContext.globalState.get<string>(COPILOT_CLI_MODEL_MEMENTO_KEY, defaultModel.id)?.trim()?.toLowerCase();
+		const preferredModelId = this.extensionContext.globalState.get<string>(COPILOT_CLI_MODEL_MEMENTO_KEY, defaultModel.id)?.trim()?.toLowerCase() ?? defaultModel.id;
 
-		return models.find(m => m.id.toLowerCase() === preferredModelId)?.id ?? defaultModel.id;
+		return models.find(m => matchesCopilotCLIModel(m, preferredModelId))?.id ?? defaultModel.id;
 	}
 
 	public async setDefaultModel(modelId: string | undefined): Promise<void> {