@dianshuv/copilot-api 0.4.2 → 0.5.0

package/README.md

@@ -61,7 +61,7 @@ copilot-api start
 | `--proxy-env` | Use proxy from environment | false |
 | `--no-history` | Disable request history UI at `/history` | false |
 | `--history-limit` | Max history entries in memory | 1000 |
-| `--no-auto-truncate` | Disable auto-truncate when exceeding limits | false |
+| `--no-auto-truncate` | Disable auto-truncate when exceeding token limits | false |
 | `--compress-tool-results` | Compress old tool results before truncating | false |
 | `--redirect-anthropic` | Force Anthropic through OpenAI translation | false |
 | `--no-rewrite-anthropic-tools` | Don't rewrite server-side tools | false |
@@ -106,6 +106,15 @@ copilot-api start
 | `/history` | GET | Request history Web UI with token analytics (enabled by default) |
 | `/history/api/*` | GET/DELETE | History API endpoints |
 
+## Auto-Truncate
+
+When enabled (default), auto-truncate automatically compacts conversation history when it exceeds the model's token limit. This prevents request failures due to context overflow.
+
+- **Token-based truncation**: Uses the model's `max_context_window_tokens` from the Copilot API to determine when truncation is needed. A 2% safety margin is applied.
+- **No preset byte limit**: There is no hardcoded request body size limit. If the Copilot API returns a 413 (Request Entity Too Large), the proxy dynamically learns the byte limit and applies it to subsequent requests.
+- **Smart compression**: With `--compress-tool-results`, old tool results are compressed before removing messages, preserving more conversation context.
+- **Orphan filtering**: After truncation, orphaned tool results (without matching tool calls) are automatically removed.
+
 ## Using with Claude Code
 
 Create `.claude/settings.json` in your project:

package/dist/main.mjs

@@ -68,7 +68,14 @@ const EDITOR_PLUGIN_VERSION = `copilot-chat/${COPILOT_VERSION}`;
 const USER_AGENT = `GitHubCopilotChat/${COPILOT_VERSION}`;
 const API_VERSION = "2025-04-01";
 const copilotBaseUrl = (state) => state.accountType === "individual" ? "https://api.githubcopilot.com" : `https://api.${state.accountType}.githubcopilot.com`;
-const copilotHeaders = (state, vision = false) => {
+function hasHeaderKey(headers, key) {
+	const lowerKey = key.toLowerCase();
+	return Object.keys(headers).some((existingKey) => {
+		return existingKey.toLowerCase() === lowerKey;
+	});
+}
+function copilotHeaders(state, visionOrOptions) {
+	const options = typeof visionOrOptions === "boolean" ? { vision: visionOrOptions } : visionOrOptions ?? {};
 	const headers = {
 		Authorization: `Bearer ${state.copilotToken}`,
 		"content-type": standardHeaders()["content-type"],
@@ -76,14 +83,15 @@ const copilotHeaders = (state, vision = false) => {
 		"editor-version": `vscode/${state.vsCodeVersion}`,
 		"editor-plugin-version": EDITOR_PLUGIN_VERSION,
 		"user-agent": USER_AGENT,
-		"openai-intent": "conversation-panel",
+		"openai-intent": options.intent ?? "conversation-panel",
 		"x-github-api-version": API_VERSION,
 		"x-request-id": randomUUID(),
 		"x-vscode-user-agent-library-version": "electron-fetch"
 	};
-	if (vision) headers["copilot-vision-request"] = "true";
+	for (const [key, value] of Object.entries(options.modelRequestHeaders ?? {})) if (!hasHeaderKey(headers, key)) headers[key] = value;
+	if (options.vision) headers["copilot-vision-request"] = "true";
 	return headers;
-};
+}
 const GITHUB_API_BASE_URL = "https://api.github.com";
 const githubHeaders = (state) => ({
 	...standardHeaders(),
@@ -106,7 +114,7 @@ const GITHUB_APP_SCOPES = ["read:user"].join(" ");
 */
 const DEFAULT_AUTO_TRUNCATE_CONFIG = {
 	safetyMarginPercent: 2,
-	maxRequestBodyBytes: 510 * 1024,
+	maxRequestBodyBytes: Infinity,
 	preserveRecentPercent: .7
 };
 /** Dynamic byte limit that adjusts based on 413 errors */
@@ -208,6 +216,10 @@ function formatRateLimitError(copilotMessage) {
 		}
 	};
 }
+function truncateForLog(text, maxLen) {
+	if (text.length <= maxLen) return text;
+	return `${text.slice(0, maxLen)}...`;
+}
 function forwardError(c, error) {
 	if (error instanceof HTTPError) {
 		if (error.status === 413) {
@@ -246,7 +258,9 @@ function forwardError(c, error) {
 			consola.warn(`HTTP 429: Rate limit exceeded`);
 			return c.json(formattedError, 429);
 		}
-		consola.error(`HTTP ${error.status}:`, errorJson);
+		let loggedError = errorJson;
+		if (typeof errorJson === "string") loggedError = errorJson.trimStart().startsWith("<") ? `[HTML ${errorJson.length} bytes]` : truncateForLog(errorJson, 200);
+		consola.error(`HTTP ${error.status}:`, loggedError);
 		return c.json({ error: {
 			message: error.responseText,
 			type: "error"
@@ -1021,7 +1035,7 @@ const patchClaude = defineCommand({
 
 //#endregion
 //#region package.json
-var version = "0.4.2";
+var version = "0.5.0";
 
 //#endregion
 //#region src/lib/adaptive-rate-limiter.ts
@@ -2465,8 +2479,10 @@ var RequestTracker = class {
 		if (update.durationMs !== void 0) request.durationMs = update.durationMs;
 		if (update.inputTokens !== void 0) request.inputTokens = update.inputTokens;
 		if (update.outputTokens !== void 0) request.outputTokens = update.outputTokens;
+		if (update.reasoningTokens !== void 0) request.reasoningTokens = update.reasoningTokens;
 		if (update.error !== void 0) request.error = update.error;
 		if (update.queuePosition !== void 0) request.queuePosition = update.queuePosition;
+		if (update.queueWaitMs !== void 0) request.queueWaitMs = update.queueWaitMs;
 		this.renderer?.onRequestUpdate(id, update);
 	}
 	/**
@@ -2481,6 +2497,7 @@ var RequestTracker = class {
 		if (usage) {
 			request.inputTokens = usage.inputTokens;
 			request.outputTokens = usage.outputTokens;
+			if (usage.reasoningTokens !== void 0) request.reasoningTokens = usage.reasoningTokens;
 		}
 		this.renderer?.onRequestComplete(request);
 		this.requests.delete(id);
@@ -3281,6 +3298,26 @@ function createTruncationResponseMarkerOpenAI(result) {
 	return `\n\n---\n[Auto-truncated: ${result.removedMessageCount} messages removed, ${result.originalTokens} → ${result.compactedTokens} tokens (${percentage}% reduction)]`;
 }
 
+//#endregion
+//#region src/lib/message-sanitizer.ts
+const startPattern = /^\s*<system-reminder>[\s\S]*?<\/system-reminder>\n*/;
+const endPatternWithNewline = /\n+<system-reminder>[\s\S]*?<\/system-reminder>\s*$/;
+const endPatternOnly = /^\s*<system-reminder>[\s\S]*?<\/system-reminder>\s*$/;
+function removeSystemReminderTags(text) {
+	let result = text;
+	let prev;
+	do {
+		prev = result;
+		result = result.replace(startPattern, "");
+	} while (result !== prev);
+	do {
+		prev = result;
+		result = result.replace(endPatternWithNewline, "");
+	} while (result !== prev);
+	result = result.replace(endPatternOnly, "");
+	return result;
+}
+
 //#endregion
 //#region src/lib/repetition-detector.ts
 /**
@@ -3409,7 +3446,10 @@ const createChatCompletions = async (payload, options) => {
 	const enableVision = payload.messages.some((x) => typeof x.content !== "string" && x.content?.some((x) => x.type === "image_url"));
 	const isAgentCall = payload.messages.some((msg) => ["assistant", "tool"].includes(msg.role));
 	const headers = {
-		...copilotHeaders(state, enableVision),
+		...copilotHeaders(state, {
+			vision: enableVision,
+			intent: isAgentCall ? "conversation-agent" : "conversation-panel"
+		}),
 		"X-Initiator": options?.initiator ?? (isAgentCall ? "agent" : "user")
 	};
 	const response = await fetch(`${copilotBaseUrl(state)}/chat/completions`, {
@@ -3464,16 +3504,18 @@ function recordErrorResponse(ctx, model, error) {
 	}, Date.now() - ctx.startTime);
 }
 /** Complete TUI tracking */
-function completeTracking(trackingId, inputTokens, outputTokens, queueWaitMs) {
+function completeTracking(trackingId, inputTokens, outputTokens, queueWaitMs, reasoningTokens) {
 	if (!trackingId) return;
 	requestTracker.updateRequest(trackingId, {
 		inputTokens,
 		outputTokens,
-		queueWaitMs
+		queueWaitMs,
+		reasoningTokens
 	});
 	requestTracker.completeRequest(trackingId, 200, {
 		inputTokens,
-		outputTokens
+		outputTokens,
+		reasoningTokens
 	});
 }
 /** Fail TUI tracking */
@@ -3521,7 +3563,7 @@ async function buildFinalPayload(payload, model) {
 	}
 	try {
 		const check = await checkNeedsCompactionOpenAI(payload, model);
-		consola.debug(`Auto-truncate check: ${check.currentTokens} tokens (limit ${check.tokenLimit}), ${Math.round(check.currentBytes / 1024)}KB (limit ${Math.round(check.byteLimit / 1024)}KB), needed: ${check.needed}${check.reason ? ` (${check.reason})` : ""}`);
+		consola.debug(`Auto-truncate check: ${check.currentTokens} tokens (limit ${check.tokenLimit}), ${Math.round(check.currentBytes / 1024)}KB (limit ${check.byteLimit === Infinity ? "unlimited" : `${Math.round(check.byteLimit / 1024)}KB`}), needed: ${check.needed}${check.reason ? ` (${check.reason})` : ""}`);
 		if (!check.needed) return {
 			finalPayload: payload,
 			truncateResult: null
@@ -3593,6 +3635,9 @@ async function logPayloadSizeInfo(payload, model) {
 
 //#endregion
 //#region src/routes/chat-completions/handler.ts
+function getReasoningTokensFromOpenAIUsage(usage) {
+	return usage?.completion_tokens_details?.reasoning_tokens;
+}
 async function handleCompletion$1(c) {
 	const originalPayload = await c.req.json();
 	consola.debug("Request payload:", JSON.stringify(originalPayload).slice(-400));
@@ -3685,12 +3730,14 @@ function handleNonStreamingResponse$1(c, originalResponse, ctx) {
 	}
 	const choice = response.choices[0];
 	const usage = response.usage;
+	const reasoningTokens = getReasoningTokensFromOpenAIUsage(usage);
 	recordResponse(ctx.historyId, {
 		success: true,
 		model: response.model,
 		usage: {
 			input_tokens: usage?.prompt_tokens ?? 0,
-			output_tokens: usage?.completion_tokens ?? 0
+			output_tokens: usage?.completion_tokens ?? 0,
+			...reasoningTokens !== void 0 ? { output_tokens_details: { reasoning_tokens: reasoningTokens } } : {}
 		},
 		stop_reason: choice.finish_reason,
 		content: buildResponseContent(choice),
@@ -3699,7 +3746,8 @@ function handleNonStreamingResponse$1(c, originalResponse, ctx) {
 	if (ctx.trackingId && usage) requestTracker.updateRequest(ctx.trackingId, {
 		inputTokens: usage.prompt_tokens,
 		outputTokens: usage.completion_tokens,
-		queueWaitMs: ctx.queueWaitMs
+		queueWaitMs: ctx.queueWaitMs,
+		reasoningTokens
 	});
 	return c.json(response);
 }
@@ -3729,6 +3777,7 @@ function createStreamAccumulator() {
 		model: "",
 		inputTokens: 0,
 		outputTokens: 0,
+		reasoningTokens: 0,
 		finishReason: "",
 		content: "",
 		toolCalls: [],
@@ -3766,7 +3815,7 @@ async function handleStreamingResponse$1(opts) {
 			await stream.writeSSE(chunk);
 		}
 		recordStreamSuccess(acc, payload.model, ctx);
-		completeTracking(ctx.trackingId, acc.inputTokens, acc.outputTokens, ctx.queueWaitMs);
+		completeTracking(ctx.trackingId, acc.inputTokens, acc.outputTokens, ctx.queueWaitMs, acc.reasoningTokens);
 	} catch (error) {
 		recordStreamError({
 			acc,
@@ -3786,6 +3835,7 @@ function parseStreamChunk(chunk, acc, checkRepetition) {
 		if (parsed.usage) {
 			acc.inputTokens = parsed.usage.prompt_tokens;
 			acc.outputTokens = parsed.usage.completion_tokens;
+			acc.reasoningTokens = getReasoningTokensFromOpenAIUsage(parsed.usage) ?? 0;
 		}
 		const choice = parsed.choices[0];
 		if (choice) {
@@ -3826,7 +3876,8 @@ function recordStreamSuccess(acc, fallbackModel, ctx) {
 		model: acc.model || fallbackModel,
 		usage: {
 			input_tokens: acc.inputTokens,
-			output_tokens: acc.outputTokens
+			output_tokens: acc.outputTokens,
+			...acc.reasoningTokens > 0 ? { output_tokens_details: { reasoning_tokens: acc.reasoningTokens } } : {}
 		},
 		stop_reason: acc.finishReason || void 0,
 		content: {
@@ -3845,7 +3896,7 @@ function convertOpenAIMessages(messages) {
 	return messages.map((msg) => {
 		const result = {
 			role: msg.role,
-			content: typeof msg.content === "string" ? msg.content : JSON.stringify(msg.content)
+			content: typeof msg.content === "string" ? removeSystemReminderTags(msg.content) : JSON.stringify(msg.content)
 		};
 		if ("tool_calls" in msg && msg.tool_calls) result.tool_calls = msg.tool_calls.map((tc) => ({
 			id: tc.id,
@@ -6006,7 +6057,10 @@ async function createAnthropicMessages(payload, options) {
 	});
 	const isAgentCall = filteredPayload.messages.some((msg) => msg.role === "assistant");
 	const headers = {
-		...copilotHeaders(state, enableVision),
+		...copilotHeaders(state, {
+			vision: enableVision,
+			intent: isAgentCall ? "conversation-agent" : "conversation-panel"
+		}),
 		"X-Initiator": options?.initiator ?? (isAgentCall ? "agent" : "user"),
 		"anthropic-version": "2023-06-01"
 	};
@@ -6141,12 +6195,12 @@ function convertAnthropicMessages(messages) {
 	return messages.map((msg) => {
 		if (typeof msg.content === "string") return {
 			role: msg.role,
-			content: msg.content
+			content: removeSystemReminderTags(msg.content)
 		};
 		const content = msg.content.map((block) => {
 			if (block.type === "text") return {
 				type: "text",
-				text: block.text
+				text: removeSystemReminderTags(block.text)
 			};
 			if (block.type === "tool_use") return {
 				type: "tool_use",
@@ -6213,9 +6267,13 @@ function createAnthropicStreamAccumulator() {
 		stopReason: "",
 		content: "",
 		toolCalls: [],
+		serverToolResults: [],
 		currentToolCall: null
 	};
 }
+function isServerToolResultType(type) {
+	return type !== "tool_result" && type.endsWith("_tool_result");
+}
 function processAnthropicEvent(event, acc) {
 	switch (event.type) {
 		case "content_block_delta":
@@ -6238,11 +6296,14 @@ function handleContentBlockDelta(delta, acc) {
 	else if (delta.type === "input_json_delta" && acc.currentToolCall) acc.currentToolCall.input += delta.partial_json;
 }
 function handleContentBlockStart(block, acc) {
-	if (block.type === "tool_use") acc.currentToolCall = {
-		id: block.id,
-		name: block.name,
-		input: ""
-	};
+	if (block.type === "tool_use") {
+		const toolBlock = block;
+		acc.currentToolCall = {
+			id: toolBlock.id,
+			name: toolBlock.name,
+			input: ""
+		};
+	} else if (isServerToolResultType(block.type)) acc.serverToolResults.push(block);
 }
 function handleContentBlockStop(acc) {
 	if (acc.currentToolCall) {
@@ -6257,6 +6318,32 @@ function handleMessageDelta(delta, usage, acc) {
 		acc.outputTokens = usage.output_tokens;
 	}
 }
+function recordAnthropicStreamingResponse(acc, fallbackModel, ctx) {
+	const contentBlocks = [];
+	if (acc.content) contentBlocks.push({
+		type: "text",
+		text: acc.content
+	});
+	for (const tc of acc.toolCalls) contentBlocks.push({
+		type: "tool_use",
+		...tc
+	});
+	for (const result of acc.serverToolResults) contentBlocks.push(result);
+	recordResponse(ctx.historyId, {
+		success: true,
+		model: acc.model || fallbackModel,
+		usage: {
+			input_tokens: acc.inputTokens,
+			output_tokens: acc.outputTokens
+		},
+		stop_reason: acc.stopReason || void 0,
+		content: contentBlocks.length > 0 ? {
+			role: "assistant",
+			content: contentBlocks
+		} : null,
+		toolCalls: acc.toolCalls.length > 0 ? acc.toolCalls : void 0
+	}, Date.now() - ctx.startTime);
+}
 
 //#endregion
 //#region src/routes/messages/non-stream-translation.ts
@@ -6738,7 +6825,7 @@ async function handleDirectAnthropicCompletion(c, anthropicPayload, ctx, initiat
 	let truncateResult;
 	if (state.autoTruncate && selectedModel) {
 		const check = await checkNeedsCompactionAnthropic(anthropicPayload, selectedModel);
-		consola.debug(`[Anthropic] Auto-truncate check: ${check.currentTokens} tokens (limit ${check.tokenLimit}), ${Math.round(check.currentBytes / 1024)}KB (limit ${Math.round(check.byteLimit / 1024)}KB), needed: ${check.needed}${check.reason ? ` (${check.reason})` : ""}`);
+		consola.debug(`[Anthropic] Auto-truncate check: ${check.currentTokens} tokens (limit ${check.tokenLimit}), ${Math.round(check.currentBytes / 1024)}KB (limit ${check.byteLimit === Infinity ? "unlimited" : `${Math.round(check.byteLimit / 1024)}KB`}), needed: ${check.needed}${check.reason ? ` (${check.reason})` : ""}`);
 		if (check.needed) try {
 			truncateResult = await autoTruncateAnthropic(anthropicPayload, selectedModel);
 			if (truncateResult.wasCompacted) effectivePayload = truncateResult.payload;
@@ -6875,7 +6962,7 @@ async function handleDirectAnthropicStreamingResponse(opts) {
 				data: rawEvent.data
 			});
 		}
-		recordStreamingResponse$1(acc, anthropicPayload.model, ctx);
+		recordAnthropicStreamingResponse(acc, anthropicPayload.model, ctx);
 		completeTracking(ctx.trackingId, acc.inputTokens, acc.outputTokens, ctx.queueWaitMs);
 	} catch (error) {
 		consola.error("Direct Anthropic stream error:", error);
@@ -6893,31 +6980,6 @@ async function handleDirectAnthropicStreamingResponse(opts) {
 		});
 	}
 }
-function recordStreamingResponse$1(acc, fallbackModel, ctx) {
-	const contentBlocks = [];
-	if (acc.content) contentBlocks.push({
-		type: "text",
-		text: acc.content
-	});
-	for (const tc of acc.toolCalls) contentBlocks.push({
-		type: "tool_use",
-		...tc
-	});
-	recordResponse(ctx.historyId, {
-		success: true,
-		model: acc.model || fallbackModel,
-		usage: {
-			input_tokens: acc.inputTokens,
-			output_tokens: acc.outputTokens
-		},
-		stop_reason: acc.stopReason || void 0,
-		content: contentBlocks.length > 0 ? {
-			role: "assistant",
-			content: contentBlocks
-		} : null,
-		toolCalls: acc.toolCalls.length > 0 ? acc.toolCalls : void 0
-	}, Date.now() - ctx.startTime);
-}
 
 //#endregion
 //#region src/routes/messages/subagent-marker.ts
@@ -7084,7 +7146,7 @@ async function handleStreamingResponse(opts) {
 			acc,
 			checkRepetition
 		});
-		recordStreamingResponse(acc, anthropicPayload.model, ctx);
+		recordAnthropicStreamingResponse(acc, anthropicPayload.model, ctx);
 		completeTracking(ctx.trackingId, acc.inputTokens, acc.outputTokens, ctx.queueWaitMs);
 	} catch (error) {
 		consola.error("Stream error:", error);
@@ -7163,31 +7225,6 @@ async function processStreamChunks(opts) {
 		}
 	}
 }
-function recordStreamingResponse(acc, fallbackModel, ctx) {
-	const contentBlocks = [];
-	if (acc.content) contentBlocks.push({
-		type: "text",
-		text: acc.content
-	});
-	for (const tc of acc.toolCalls) contentBlocks.push({
-		type: "tool_use",
-		...tc
-	});
-	recordResponse(ctx.historyId, {
-		success: true,
-		model: acc.model || fallbackModel,
-		usage: {
-			input_tokens: acc.inputTokens,
-			output_tokens: acc.outputTokens
-		},
-		stop_reason: acc.stopReason || void 0,
-		content: contentBlocks.length > 0 ? {
-			role: "assistant",
-			content: contentBlocks
-		} : null,
-		toolCalls: acc.toolCalls.length > 0 ? acc.toolCalls : void 0
-	}, Date.now() - ctx.startTime);
-}
 
 //#endregion
 //#region src/routes/messages/handler.ts
@@ -7361,7 +7398,7 @@ modelRoutes.get("/", async (c) => {
 const createResponses = async (payload, { vision, initiator }) => {
 	if (!state.copilotToken) throw new Error("Copilot token not found");
 	const headers = {
-		...copilotHeaders(state, vision),
+		...copilotHeaders(state, { vision }),
 		"X-Initiator": initiator
 	};
 	payload.service_tier = null;
@@ -7633,7 +7670,7 @@ const handleResponses = async (c) => {
 					if (finalResult) {
 						recordResponseResult(finalResult, model, historyId, startTime);
 						const usage = finalResult.usage;
-						completeTracking(trackingId, usage?.input_tokens ?? 0, usage?.output_tokens ?? 0, queueWaitMs);
+						completeTracking(trackingId, usage?.input_tokens ?? 0, usage?.output_tokens ?? 0, queueWaitMs, usage?.output_tokens_details?.reasoning_tokens);
 					} else if (streamErrorMessage) {
 						recordResponse(historyId, {
 							success: false,
@@ -7662,7 +7699,7 @@ const handleResponses = async (c) => {
 		const result = response;
 		const usage = result.usage;
 		recordResponseResult(result, model, historyId, startTime);
-		completeTracking(trackingId, usage?.input_tokens ?? 0, usage?.output_tokens ?? 0, ctx.queueWaitMs);
+		completeTracking(trackingId, usage?.input_tokens ?? 0, usage?.output_tokens ?? 0, ctx.queueWaitMs, usage?.output_tokens_details?.reasoning_tokens);
 		consola.debug("Forwarding native Responses result:", JSON.stringify(result).slice(-400));
 		return c.json(result);
 	} catch (error) {
@@ -7713,7 +7750,8 @@ function recordResponseResult(result, fallbackModel, historyId, startTime) {
 		model: result.model || fallbackModel,
 		usage: {
 			input_tokens: usage?.input_tokens ?? 0,
-			output_tokens: usage?.output_tokens ?? 0
+			output_tokens: usage?.output_tokens ?? 0,
+			...usage?.output_tokens_details ? { output_tokens_details: { reasoning_tokens: usage.output_tokens_details.reasoning_tokens } } : {}
 		},
 		stop_reason: extractResponseStopReason(result),
 		content,
@@ -7788,6 +7826,18 @@ server.route("/history", historyRoutes);
 
 //#endregion
 //#region src/start.ts
+const VALID_ACCOUNT_TYPES = [
+	"individual",
+	"business",
+	"enterprise"
+];
+function isValidAccountType(accountType) {
+	return VALID_ACCOUNT_TYPES.includes(accountType);
+}
+function validateAccountType(accountType) {
+	if (isValidAccountType(accountType)) return;
+	throw new Error(`Invalid account type: "${accountType}". Available: ${VALID_ACCOUNT_TYPES.join(", ")}`);
+}
 /** Format limit values as "Xk" or "?" if not available */
 function formatLimit(value) {
 	return value ? `${Math.round(value / 1e3)}k` : "?";
@@ -7810,6 +7860,12 @@ async function runServer(options) {
 		state.verbose = true;
 	}
 	state.accountType = options.accountType;
+	try {
+		validateAccountType(state.accountType);
+	} catch (error) {
+		consola.error(error instanceof Error ? error.message : String(error));
+		process.exit(1);
+	}
 	if (options.accountType !== "individual") consola.info(`Using ${options.accountType} plan GitHub account`);
 	state.manualApprove = options.manual;
 	state.showToken = options.showToken;
@@ -7843,7 +7899,13 @@ async function runServer(options) {
 		consola.info("Using provided GitHub token");
 	} else await setupGitHubToken();
 	await setupCopilotToken();
-	await cacheModels();
+	try {
+		await cacheModels();
+	} catch (error) {
+		consola.error(`Failed to fetch available models for account type "${state.accountType}". Check that the account type matches your Copilot plan.`);
+		consola.error(error instanceof Error ? error.message : String(error));
+		process.exit(1);
+	}
 	consola.info(`Available models:\n${state.models?.data.map((m) => formatModelInfo(m)).join("\n")}`);
 	const serverUrl = `http://${options.host ?? "localhost"}:${options.port}`;
 	if (options.claudeCode) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@dianshuv/copilot-api",
-  "version": "0.4.2",
+  "version": "0.5.0",
   "description": "Turn GitHub Copilot into OpenAI/Anthropic API compatible server. Usable with Claude Code!",
   "author": "dianshuv",
   "type": "module",