[pull] main from microsoft:main

.github/ISSUE_TEMPLATE/team/testplan_item.md

@@ -0,0 +1,18 @@
+---
+name: Test Plan Item
+about: Create test plan item for endgame
+title: ""
+labels: testplan-item
+assignees: ""
+---
+
+Refs: <!-- Refer to the issue that this test plan item is testing. -->
+
+- [ ] anyOS
+- [ ] anyOS
+
+Complexity: 4
+
+---
+
+<!-- Please write your test here. -->

build/lib/screenshotDiffReport.ts

@@ -32,7 +32,7 @@ const __dirname = path.dirname(fileURLToPath(import.meta.url));
 const COMMENT_MARKER = '<!-- screenshot-diff-report -->';
 const EXPAND_FIRST_N = 5;
 const EXCLUDED_LABELS = new Set(['animated', 'flaky']);
-const MAX_BODY_BYTES = 750 * 1024;
+const MAX_BODY_BYTES = 300 * 1024;
 
 // ---------------------------------------------------------------------------
 // Pixel-level image comparison

extensions/copilot/.vscodeignore

@@ -30,6 +30,7 @@ node_modules/@github/copilot/sharp/**
 !node_modules/@github/copilot/sdk/worker/*.js
 !node_modules/@github/copilot/sdk/sharp/**
 !node_modules/@github/copilot/sdk/definitions/*.yaml
+!node_modules/@github/copilot/sdk/definitions/sidekick/*.yaml
 !CHANGELOG.md
 !README.md
 !package.json

extensions/copilot/package.json

@@ -1951,6 +1951,21 @@
 										"default": false,
 										"markdownDescription": "Whether Zero Data Retention (ZDR) is enabled for this endpoint. When `true`, `previous_response_id` will not be sent in requests via Responses API."
 									},
+									"supportsReasoningEffort": {
+										"type": "array",
+										"markdownDescription": "Reasoning effort levels the model accepts (e.g. `[\"low\", \"medium\", \"high\"]`). When set, a `Thinking Effort` picker is shown in the model picker and the chosen value is forwarded to the model. Levels supported by mainstream OpenAI-compatible servers are `minimal`, `low`, `medium`, `high`.",
+										"items": {
+											"type": "string"
+										}
+									},
+									"reasoningEffortFormat": {
+										"type": "string",
+										"enum": [
+											"chat-completions",
+											"responses"
+										],
+										"markdownDescription": "Body shape used to forward the reasoning effort to the model. `chat-completions` sends a top-level `reasoning_effort` string. `responses` sends a nested `reasoning.effort` object. When unset the format follows the URL: `/responses` \u2192 nested, otherwise top-level."
+									},
 									"requestHeaders": {
 										"type": "object",
 										"description": "Additional HTTP headers to include with requests to this model. These reserved headers are not allowed and ignored if present: forbidden request headers (https://developer.mozilla.org/en-US/docs/Glossary/Forbidden_request_header), forwarding headers ('forwarded', 'x-forwarded-for', 'x-forwarded-host', 'x-forwarded-proto'), and others ('api-key', 'authorization', 'content-type', 'openai-intent', 'x-github-api-version', 'x-initiator', 'x-interaction-id', 'x-interaction-type', 'x-onbehalf-extension-id', 'x-request-id', 'x-vscode-user-agent-library-version'). Pattern-based forbidden headers ('proxy-*', 'sec-*', 'x-http-method*' with forbidden methods) are also blocked.",
@@ -2050,6 +2065,21 @@
 										"default": false,
 										"markdownDescription": "Whether Zero Data Retention (ZDR) is enabled for this endpoint. When `true`, `previous_response_id` will not be sent in requests via Responses API."
 									},
+									"supportsReasoningEffort": {
+										"type": "array",
+										"markdownDescription": "Reasoning effort levels the model accepts (e.g. `[\"low\", \"medium\", \"high\"]`). When set, a `Thinking Effort` picker is shown in the model picker and the chosen value is forwarded to the model. Levels supported by mainstream OpenAI-compatible servers are `minimal`, `low`, `medium`, `high`.",
+										"items": {
+											"type": "string"
+										}
+									},
+									"reasoningEffortFormat": {
+										"type": "string",
+										"enum": [
+											"chat-completions",
+											"responses"
+										],
+										"markdownDescription": "Body shape used to forward the reasoning effort to the model. `chat-completions` sends a top-level `reasoning_effort` string. `responses` sends a nested `reasoning.effort` object. When unset the format follows the URL: `/responses` \u2192 nested, otherwise top-level."
+									},
 									"requestHeaders": {
 										"type": "object",
 										"description": "Additional HTTP headers to include with requests to this model. These reserved headers are not allowed and ignored if present: forbidden request headers (https://developer.mozilla.org/en-US/docs/Glossary/Forbidden_request_header), forwarding headers ('forwarded', 'x-forwarded-for', 'x-forwarded-host', 'x-forwarded-proto'), and others ('api-key', 'authorization', 'content-type', 'openai-intent', 'x-github-api-version', 'x-initiator', 'x-interaction-id', 'x-interaction-type', 'x-onbehalf-extension-id', 'x-request-id', 'x-vscode-user-agent-library-version'). Pattern-based forbidden headers ('proxy-*', 'sec-*', 'x-http-method*' with forbidden methods) are also blocked.",
@@ -4782,7 +4812,7 @@
 					},
 					"github.copilot.chat.cli.sessionController.enabled": {
 						"type": "boolean",
-						"default": true,
+						"default": false,
 						"markdownDescription": "%github.copilot.config.cli.sessionController.enabled%",
 						"tags": [
 							"advanced"

extensions/copilot/src/extension/byok/common/byokProvider.ts

@@ -61,6 +61,13 @@ export interface BYOKModelCapabilities {
 	supportedEndpoints?: ModelSupportedEndpoint[];
 	zeroDataRetentionEnabled?: boolean;
 	supportsReasoningEffort?: string[];
+	/**
+	 * Override the body shape used to forward the reasoning effort to the model.
+	 * - `'chat-completions'`: top-level `reasoning_effort` (default for `/chat/completions`).
+	 * - `'responses'`: nested `reasoning.effort` (default for `/responses`).
+	 * If unset the format is inferred from whether the endpoint uses the Responses API.
+	 */
+	reasoningEffortFormat?: 'chat-completions' | 'responses';
 }
 
 export interface BYOKModelRegistry {
@@ -108,7 +115,8 @@ export function resolveModelInfo(modelId: string, providerName: string, knownMod
 				tool_calls: !!knownModelInfo?.toolCalling,
 				vision: !!knownModelInfo?.vision,
 				thinking: !!knownModelInfo?.thinking,
-				adaptive_thinking: !!knownModelInfo?.adaptiveThinking
+				adaptive_thinking: !!knownModelInfo?.adaptiveThinking,
+				reasoning_effort: knownModelInfo?.supportsReasoningEffort
 			},
 			tokenizer: TokenizerType.O200K,
 			limits: {
@@ -121,7 +129,8 @@ export function resolveModelInfo(modelId: string, providerName: string, knownMod
 		is_chat_fallback: false,
 		model_picker_enabled: true,
 		supported_endpoints: knownModelInfo?.supportedEndpoints,
-		zeroDataRetentionEnabled: knownModelInfo?.zeroDataRetentionEnabled
+		zeroDataRetentionEnabled: knownModelInfo?.zeroDataRetentionEnabled,
+		reasoningEffortFormat: knownModelInfo?.reasoningEffortFormat
 	};
 	if (knownModelInfo?.requestHeaders && Object.keys(knownModelInfo.requestHeaders).length > 0) {
 		modelInfo.requestHeaders = { ...knownModelInfo.requestHeaders };

extensions/copilot/src/extension/byok/common/test/byokProvider.spec.ts

@@ -4,7 +4,7 @@
  *--------------------------------------------------------------------------------------------*/
 
 import { describe, expect, it } from 'vitest';
-import { byokKnownModelToAPIInfo, BYOKModelCapabilities } from '../byokProvider';
+import { byokKnownModelToAPIInfo, BYOKModelCapabilities, resolveModelInfo } from '../byokProvider';
 
 describe('byokKnownModelToAPIInfo', () => {
 	const baseCapabilities: BYOKModelCapabilities = {
@@ -43,3 +43,31 @@ describe('byokKnownModelToAPIInfo', () => {
 		expect(info.capabilities.editTools).toBeUndefined();
 	});
 });
+
+describe('resolveModelInfo', () => {
+	const baseCapabilities: BYOKModelCapabilities = {
+		name: 'TestModel',
+		maxInputTokens: 1000,
+		maxOutputTokens: 100,
+		toolCalling: true,
+		vision: false,
+	};
+
+	it('propagates supportsReasoningEffort and reasoningEffortFormat from BYOK capabilities into chat-endpoint inputs', () => {
+		const info = resolveModelInfo('m1', 'TestProvider', undefined, {
+			...baseCapabilities,
+			supportsReasoningEffort: ['low', 'medium', 'high'],
+			reasoningEffortFormat: 'responses',
+		});
+
+		expect(info.capabilities.supports.reasoning_effort).toEqual(['low', 'medium', 'high']);
+		expect(info.reasoningEffortFormat).toBe('responses');
+	});
+
+	it('omits the reasoning effort capability when the model does not declare it', () => {
+		const info = resolveModelInfo('m1', 'TestProvider', undefined, baseCapabilities);
+
+		expect(info.capabilities.supports.reasoning_effort).toBeUndefined();
+		expect(info.reasoningEffortFormat).toBeUndefined();
+	});
+});

extensions/copilot/src/extension/byok/node/openAIEndpoint.ts

@@ -5,7 +5,7 @@
 import type { CancellationToken } from 'vscode';
 import { IChatMLFetcher } from '../../../platform/chat/common/chatMLFetcher';
 import { ChatFetchResponseType, ChatResponse } from '../../../platform/chat/common/commonTypes';
-import { IConfigurationService } from '../../../platform/configuration/common/configurationService';
+import { ConfigKey, IConfigurationService } from '../../../platform/configuration/common/configurationService';
 import { IDomainService } from '../../../platform/endpoint/common/domainService';
 import { IChatModelInformation } from '../../../platform/endpoint/common/endpointProvider';
 import { ChatEndpoint } from '../../../platform/endpoint/node/chatEndpoint';
@@ -250,6 +250,7 @@ export class OpenAIEndpoint extends ChatEndpoint {
 				// Don't use a response ID from CAPI or when zero data retention is enabled
 				body.previous_response_id = undefined;
 			}
+			this._applyReasoningEffort(body, options);
 			return body;
 		} else if (this.useMessagesApi) {
 			// Delegate to base ChatEndpoint for Messages API dispatch
@@ -263,10 +264,44 @@ export class OpenAIEndpoint extends ChatEndpoint {
 				}
 			};
 			const body = createCapiRequestBody(options, this.model, callback);
+			this._applyReasoningEffort(body, options);
 			return body;
 		}
 	}
 
+	/**
+	 * Forwards the per-request reasoning effort to the model body in the shape the endpoint expects.
+	 * Default shape mirrors the API path (`Responses` \u2192 nested `reasoning.effort`, `Chat Completions` \u2192 top-level `reasoning_effort`).
+	 * `IChatModelInformation.reasoningEffortFormat` overrides the default so users hosting OpenAI-compatible servers
+	 * with diverging conventions (e.g. nested `reasoning.effort` on `/chat/completions`) can opt in deterministically.
+	 */
+	private _applyReasoningEffort(body: IEndpointBody, options: ICreateEndpointBodyOptions): void {
+		const supports = this.supportsReasoningEffort;
+		if (!supports?.length) {
+			return;
+		}
+		const format = this.modelMetadata.reasoningEffortFormat
+			?? (this.useResponsesApi ? 'responses' : 'chat-completions');
+		const override = this._configurationService.getConfig(ConfigKey.Advanced.ReasoningEffortOverride);
+		const requested = override || options.modelCapabilities?.reasoningEffort || body.reasoning?.effort || body.reasoning_effort;
+		const effort = requested && supports.includes(requested) ? requested : undefined;
+		// Scrub any pre-populated effort first so unsupported values (e.g. the hard-coded `medium` default
+		// from `createResponsesRequestBody`) cannot leak through, then write the resolved value into the
+		// expected shape.
+		if (body.reasoning) {
+			const { effort: _drop, ...rest } = body.reasoning;
+			body.reasoning = Object.keys(rest).length > 0 ? rest : undefined;
+		}
+		body.reasoning_effort = undefined;
+		if (effort) {
+			if (format === 'responses') {
+				body.reasoning = { ...body.reasoning, effort };
+			} else {
+				body.reasoning_effort = effort;
+			}
+		}
+	}
+
 	override interceptBody(body: IEndpointBody | undefined): void {
 		super.interceptBody(body);
 		// TODO @lramos15 - We should do this for all models and not just here