@f5xc-salesdemos/xcsh 17.1.0 → 17.1.2

package/package.json

@@ -1,7 +1,7 @@
 {
 	"type": "module",
 	"name": "@f5xc-salesdemos/xcsh",
-	"version": "17.1.0",
+	"version": "17.1.2",
 	"description": "Coding agent CLI with read, bash, edit, write tools and session management",
 	"homepage": "https://github.com/f5xc-salesdemos/xcsh",
 	"author": "Can Boluk",
@@ -46,12 +46,12 @@
 	"dependencies": {
 		"@agentclientprotocol/sdk": "0.16.1",
 		"@mozilla/readability": "^0.6",
-		"@f5xc-salesdemos/xcsh-stats": "17.1.0",
-		"@f5xc-salesdemos/pi-agent-core": "17.1.0",
-		"@f5xc-salesdemos/pi-ai": "17.1.0",
-		"@f5xc-salesdemos/pi-natives": "17.1.0",
-		"@f5xc-salesdemos/pi-tui": "17.1.0",
-		"@f5xc-salesdemos/pi-utils": "17.1.0",
+		"@f5xc-salesdemos/xcsh-stats": "17.1.2",
+		"@f5xc-salesdemos/pi-agent-core": "17.1.2",
+		"@f5xc-salesdemos/pi-ai": "17.1.2",
+		"@f5xc-salesdemos/pi-natives": "17.1.2",
+		"@f5xc-salesdemos/pi-tui": "17.1.2",
+		"@f5xc-salesdemos/pi-utils": "17.1.2",
 		"@sinclair/typebox": "^0.34",
 		"@xterm/headless": "^6.0",
 		"ajv": "^8.18",

package/src/config/model-registry.ts

@@ -247,7 +247,12 @@ const ModelOverrideSchema = Type.Object({
 type ModelOverride = Static<typeof ModelOverrideSchema>;
 
 const ProviderDiscoverySchema = Type.Object({
-	type: Type.Union([Type.Literal("ollama"), Type.Literal("llama.cpp"), Type.Literal("lm-studio")]),
+	type: Type.Union([
+		Type.Literal("ollama"),
+		Type.Literal("llama.cpp"),
+		Type.Literal("lm-studio"),
+		Type.Literal("openai-compat"),
+	]),
 });
 
 const ProviderAuthSchema = Type.Union([Type.Literal("apiKey"), Type.Literal("none")]);
@@ -831,6 +836,29 @@ export class ModelRegistry {
 		this.#backgroundRefresh = refreshPromise;
 	}
 
+	/**
+	 * Await the in-flight background refresh if one is running.
+	 * Returns immediately if no background refresh is in progress.
+	 */
+	async awaitBackgroundRefresh(): Promise<void> {
+		if (this.#backgroundRefresh) {
+			await this.#backgroundRefresh;
+		}
+	}
+
+	/**
+	 * Check if any non-optional discoverable provider has no cached models yet.
+	 * Returns true on first run when the model cache is empty.
+	 */
+	hasUncachedDiscoverableProviders(): boolean {
+		for (const [, state] of this.#providerDiscoveryStates) {
+			if (state.status === "idle" && !state.optional) {
+				return true;
+			}
+		}
+		return false;
+	}
+
 	async refreshProvider(providerId: string, strategy: ModelRefreshStrategy = "online"): Promise<void> {
 		this.#reloadStaticModels();
 		for (const selector of this.#suppressedSelectors.keys()) {
@@ -1285,6 +1313,8 @@ export class ModelRegistry {
 				return this.#discoverLlamaCppModels(providerConfig);
 			case "lm-studio":
 				return this.#discoverLmStudioModels(providerConfig);
+			case "openai-compat":
+				return this.#discoverOpenAICompatModels(providerConfig);
 		}
 	}
 
@@ -1646,6 +1676,53 @@ export class ModelRegistry {
 		return this.#applyProviderModelOverrides(providerConfig.provider, discovered);
 	}
 
+	async #discoverOpenAICompatModels(providerConfig: DiscoveryProviderConfig): Promise<Model<Api>[]> {
+		const baseUrl = (providerConfig.baseUrl ?? "").replace(/\/+$/, "");
+		if (!baseUrl) {
+			throw new Error("openai-compat discovery requires a baseUrl");
+		}
+
+		const headers: Record<string, string> = { ...(providerConfig.headers ?? {}) };
+		const apiKey =
+			this.#customProviderApiKeys.get(providerConfig.provider) ??
+			(await this.#peekApiKeyForProvider(providerConfig.provider));
+		if (apiKey && apiKey !== DEFAULT_LOCAL_TOKEN && apiKey !== kNoAuth) {
+			headers.Authorization = `Bearer ${apiKey}`;
+		}
+
+		const modelsUrl = `${baseUrl}/models`;
+		const response = await fetch(modelsUrl, {
+			headers: { Accept: "application/json", ...headers },
+			signal: AbortSignal.timeout(3000),
+		});
+		if (!response.ok) {
+			throw new Error(`HTTP ${response.status} from ${modelsUrl}`);
+		}
+		const payload = (await response.json()) as { data?: Array<{ id: string }> };
+		const items = payload.data ?? [];
+		const discovered: Model<Api>[] = [];
+		for (const item of items) {
+			const id = item.id;
+			if (!id) continue;
+			discovered.push(
+				enrichModelThinking({
+					id,
+					name: id,
+					api: providerConfig.api,
+					provider: providerConfig.provider,
+					baseUrl,
+					reasoning: false,
+					input: ["text"],
+					cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
+					contextWindow: 128000,
+					maxTokens: 8192,
+					headers,
+				}),
+			);
+		}
+		return this.#applyProviderModelOverrides(providerConfig.provider, discovered);
+	}
+
 	#normalizeLlamaCppBaseUrl(baseUrl?: string): string {
 		const defaultBaseUrl = "http://127.0.0.1:8080";
 		const raw = baseUrl || defaultBaseUrl;

package/src/main.ts

@@ -624,6 +624,19 @@ export async function runRootCommand(parsed: Args, rawArgs: string[]): Promise<v
 
 	const cwd = getProjectDir();
 	await logger.time("settings:init", Settings.init, { cwd });
+
+	// F5 XC profile loading — optional, never blocks startup.
+	// NOTE: This runs in the CLI path only. SDK consumers using createAgentSession()
+	// directly must call ProfileService.init(configDir).loadActive() themselves.
+	try {
+		const { ProfileService } = await import("./services/f5xc-profile");
+		const { getF5XCConfigDir } = await import("@f5xc-salesdemos/pi-utils");
+		const profileService = ProfileService.init(getF5XCConfigDir());
+		await profileService.loadActive();
+	} catch {
+		// F5 XC auth is optional — silently continue if anything fails
+	}
+
 	if (parsedArgs.mode === "rpc") {
 		applyRpcDefaultSettingOverrides();
 	}

package/src/sdk.ts

@@ -23,11 +23,11 @@ import {
 	prompt,
 	Snowflake,
 } from "@f5xc-salesdemos/pi-utils";
-import chalk from "chalk";
 import { AsyncJobManager, isBackgroundJobSupportEnabled } from "./async";
 import { createAutoresearchExtension } from "./autoresearch";
 import { loadCapability } from "./capability";
 import { type Rule, ruleCapability } from "./capability/rule";
+import { hasLiteLLMEnv } from "./config/auto-config";
 import { ModelRegistry } from "./config/model-registry";
 import { formatModelString, parseModelPattern, parseModelString, resolveModelRoleValue } from "./config/model-resolver";
 import { loadPromptTemplates as loadPromptTemplatesInternal, type PromptTemplate } from "./config/prompt-templates";
@@ -780,6 +780,13 @@ export async function createAgentSession(options: CreateAgentSessionOptions = {}
 	const modelMatchPreferences = {
 		usageOrder: settings.getStorage()?.getModelUsageOrder(),
 	};
+	// When LiteLLM is configured and no model cache exists yet (first run),
+	// await the background refresh so model discovery from the proxy completes
+	// before we select a default model. Bounded by the 3s probe timeout.
+	if (!options.modelRegistry && hasLiteLLMEnv() && modelRegistry.hasUncachedDiscoverableProviders()) {
+		await logger.time("awaitLiteLLMDiscovery", () => modelRegistry.awaitBackgroundRefresh());
+	}
+
 	const defaultRoleSpec = logger.time("resolveDefaultModelRole", () =>
 		resolveModelRoleValue(settings.getModelRole("default"), modelRegistry.getAvailable(), {
 			settings,
@@ -1069,8 +1076,8 @@ export async function createAgentSession(options: CreateAgentSessionOptions = {}
 		if (enableMCP) {
 			const mcpResult = await logger.time("discoverAndLoadMCPTools", discoverAndLoadMCPTools, cwd, {
 				onConnecting: serverNames => {
-					if (options.hasUI && serverNames.length > 0) {
-						process.stderr.write(`${chalk.gray(`Connecting to MCP servers: ${serverNames.join(", ")}…`)}\n`);
+					if (serverNames.length > 0) {
+						logger.debug("Connecting to MCP servers", { servers: serverNames });
 					}
 				},
 				enableProjectConfig: settings.get("mcp.enableProjectConfig") ?? true,