pi-free 2.2.7 → 2.2.8

package/lib/model-detection.ts

@@ -4,7 +4,7 @@
  * Used for failover when providers hit rate limits.
  */
 
-import type { Model } from "@earendil-works/pi-ai";
+import type { Model } from "@earendil-works/pi-ai/compat";
 import type { ProviderModelConfig } from "./types.ts";
 
 export interface ModelInfo {

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "pi-free",
-	"version": "2.2.7",
+	"version": "2.2.8",
 	"type": "module",
 	"description": "AI model providers for Pi with free model filtering and dynamic model fetching",
 	"keywords": [
@@ -56,14 +56,14 @@
 		"smoke:openmodel": "tsx scripts/smoke-openmodel-wire-format.ts"
 	},
 	"peerDependencies": {
-		"@earendil-works/pi-ai": ">=0.80.0",
-		"@earendil-works/pi-coding-agent": ">=0.80.0",
-		"@earendil-works/pi-tui": ">=0.80.0"
+		"@earendil-works/pi-ai": ">=0.80.8",
+		"@earendil-works/pi-coding-agent": ">=0.80.8",
+		"@earendil-works/pi-tui": ">=0.80.8"
 	},
 	"devDependencies": {
 		"@vitest/ui": "^4.1.10",
 		"tsx": "^4.23.0",
-		"typescript": "^6.0.3",
+		"typescript": "^7.0.2",
 		"vitest": "^4.1.10"
 	},
 	"pi": {

package/provider-helper.ts

@@ -24,6 +24,33 @@ import { enhanceModelNameWithCodingIndex } from "./provider-failover/benchmark-l
 
 const _logger = createLogger("provider-helper");
 
+interface ContextWithOAuthStatus {
+	model?: unknown;
+	modelRegistry?: {
+		isUsingOAuth?: (model: unknown) => boolean;
+	};
+}
+
+/**
+ * Check OAuth status through Pi's ModelRegistry compatibility facade.
+ * Newer Pi versions no longer expose authStorage on the context.
+ */
+export function isCurrentModelOAuth(ctx: unknown): boolean {
+	if (!ctx || typeof ctx !== "object") return false;
+	const context = ctx as ContextWithOAuthStatus;
+	return Boolean(
+		context.model && context.modelRegistry?.isUsingOAuth?.(context.model),
+	);
+}
+
+export function isOAuthCredential(
+	credential: unknown,
+): credential is { type: "oauth"; access: string } {
+	if (!credential || typeof credential !== "object") return false;
+	const candidate = credential as { type?: unknown; access?: unknown };
+	return candidate.type === "oauth" && typeof candidate.access === "string";
+}
+
 // =============================================================================
 // Types
 // =============================================================================
@@ -124,11 +151,7 @@ export async function loadCachedOrFetchModels(
 	const ttlMs = options?.ttlMs ?? DEFAULT_PROVIDER_CACHE_TTL_MS;
 	const cached = loadProviderCache(providerId);
 
-	if (
-		cached &&
-		cached.length > 0 &&
-		isProviderCacheFresh(providerId, ttlMs)
-	) {
+	if (cached && cached.length > 0 && isProviderCacheFresh(providerId, ttlMs)) {
 		return cached;
 	}
 
@@ -336,8 +359,7 @@ export function setupProvider(
 			if (tosShown || ctx.model?.provider !== providerId) return;
 			tosShown = true;
 			if (config.hasKey) return;
-			const cred = ctx.modelRegistry.authStorage.get(providerId);
-			if (cred?.type === "oauth") return;
+			if (isCurrentModelOAuth(ctx)) return;
 			ctx.ui.notify(
 				`Using ${providerId} free models. Set API key for paid access. Terms: ${tosUrl}`,
 				"info",

package/providers/anyapi/anyapi.ts

@@ -1,270 +1,270 @@
-/**
- * AnyAPI provider extension.
- *
- * AnyAPI is an OpenAI-compatible gateway with a free plan and a catalog of
- * explicitly free models. It exposes the catalog at /v1/models and routes
- * chat requests through /v1/chat/completions.
- *
- * Setup:
- *   ANYAPI_API_KEY=...
- *   # or add anyapi_api_key to ~/.pi/free.json
- */
-
-import type {
-	ExtensionAPI,
-	ProviderModelConfig,
-} from "@earendil-works/pi-coding-agent";
-import {
-	applyHidden,
-	getAnyapiApiKey,
-	getAnyapiShowPaid,
-} from "../../config.ts";
-import {
-	BASE_URL_ANYAPI,
-	DEFAULT_FETCH_TIMEOUT_MS,
-	PROVIDER_ANYAPI,
-} from "../../constants.ts";
-import { createLogger } from "../../lib/logger.ts";
-import { loadProviderCache } from "../../lib/provider-cache.ts";
-import { isFreeModel, registerWithGlobalToggle } from "../../lib/registry.ts";
-import { safeEnrichModelsWithModelsDev } from "../../lib/model-metadata.ts";
-import { fetchWithRetry, mapOpenRouterModel } from "../../lib/util.ts";
-import {
-	createReRegister,
-	loadCachedOrFetchModels,
-	setupProvider,
-} from "../../provider-helper.ts";
-
-const _logger = createLogger("anyapi");
-
-interface AnyApiModel {
-	id: string;
-	name?: string;
-	context_length?: number;
-	max_completion_tokens?: number | null;
-	top_provider?: {
-		context_length?: number | null;
-		max_completion_tokens?: number | null;
-	};
-	pricing?: {
-		prompt?: string | number | null;
-		completion?: string | number | null;
-		input_cache_read?: string | number | null;
-		input_cache_write?: string | number | null;
-	};
-	architecture?: {
-		input_modalities?: string[] | null;
-		output_modalities?: string[] | null;
-	};
-	supported_parameters?: string[] | null;
-	tags?: string[];
-	isFree?: boolean;
-}
-
-const ANYAPI_METADATA_VERSION = 1;
-
-type AnyApiProviderModel = ProviderModelConfig & {
-	_pricingKnown?: boolean;
-	_freeKnown?: boolean;
-	_isFree?: boolean;
-	_anyapiMetadataVersion?: number;
-};
-
-function hasPricing(model: AnyApiModel): boolean {
-	return (
-		(model.pricing?.prompt !== null && model.pricing?.prompt !== undefined) ||
-		(model.pricing?.completion !== null &&
-			model.pricing?.completion !== undefined) ||
-		(model.pricing?.input_cache_read !== null &&
-			model.pricing?.input_cache_read !== undefined) ||
-		(model.pricing?.input_cache_write !== null &&
-			model.pricing?.input_cache_write !== undefined)
-	);
-}
-
-function normalizePrice(
-	value: string | number | null | undefined,
-): string | null {
-	return value === null || value === undefined ? null : String(value);
-}
-
-/**
- * Detect AnyAPI's explicitly free model labels without treating every model
- * with omitted pricing as free. The API may also expose an authoritative flag
- * or zero pricing, both of which are handled here.
- */
-export function isAnyApiFreeModel(model: AnyApiModel): boolean {
-	if (typeof model.isFree === "boolean") return model.isFree;
-
-	const label = `${model.id} ${model.name ?? ""}`.toLowerCase();
-	if (/\bfree\b/.test(label)) return true;
-
-	if (!hasPricing(model)) return false;
-	const input = Number(model.pricing?.prompt);
-	const output = Number(model.pricing?.completion);
-	return input === 0 && output === 0;
-}
-
-export function mapAnyApiModel(model: AnyApiModel): AnyApiProviderModel {
-	const name = model.name ?? model.id;
-	const pricingKnown = hasPricing(model);
-	const freeKnown =
-		typeof model.isFree === "boolean" ||
-		/\bfree\b/i.test(`${model.id} ${name}`) ||
-		(pricingKnown && isAnyApiFreeModel(model));
-
-	const tags = model.tags ?? [];
-	const supportsReasoning =
-		tags.includes("reasoning") || tags.includes("chat_completions:reasoning");
-	const supportsVision =
-		tags.includes("vision") || tags.includes("chat_completions:vision");
-
-	const mapped = mapOpenRouterModel({
-		...model,
-		name,
-		supported_parameters:
-			model.supported_parameters ?? (supportsReasoning ? ["reasoning"] : []),
-		architecture: model.architecture ?? {
-			input_modalities: supportsVision ? ["text", "image"] : ["text"],
-			output_modalities: ["text"],
-		},
-		pricing: model.pricing
-			? {
-					prompt: normalizePrice(model.pricing.prompt),
-					completion: normalizePrice(model.pricing.completion),
-					input_cache_read: normalizePrice(model.pricing.input_cache_read),
-					input_cache_write: normalizePrice(model.pricing.input_cache_write),
-				}
-			: undefined,
-	});
-
-	return {
-		...mapped,
-		_pricingKnown: pricingKnown,
-		...(freeKnown && {
-			_freeKnown: true,
-			_isFree: isAnyApiFreeModel(model),
-		}),
-	};
-}
-
-function isTextModel(model: AnyApiModel): boolean {
-	const outputModalities = model.architecture?.output_modalities ?? [];
-	if (outputModalities.length > 0 && !outputModalities.includes("text")) {
-		return false;
-	}
-
-	const tags = model.tags;
-	if (!tags || tags.length === 0) return true;
-	return tags.some((tag) => tag.startsWith("chat_completions:"));
-}
-
-async function fetchAnyApiModels(
-	apiKey: string,
-): Promise<AnyApiProviderModel[]> {
-	const response = await fetchWithRetry(
-		`${BASE_URL_ANYAPI}/models`,
-		{
-			headers: {
-				Authorization: `Bearer ${apiKey}`,
-				Accept: "application/json",
-				"Content-Type": "application/json",
-			},
-		},
-		3,
-		1000,
-		DEFAULT_FETCH_TIMEOUT_MS,
-	);
-
-	if (!response.ok) {
-		throw new Error(
-			`AnyAPI API error: ${response.status} ${response.statusText}`,
-		);
-	}
-
-	const json = (await response.json()) as { data?: AnyApiModel[] };
-	const models = (json.data ?? []).flatMap((model) => {
-		if (!model.id || !isTextModel(model)) return [];
-		return [mapAnyApiModel(model)];
-	});
-
-	_logger.info(`[anyapi] Fetched ${models.length} text models`);
-
-	// AnyAPI's /models response omits context and output limits for its
-	// catalog. Use the global models.dev catalog so canonical model IDs such
-	// as qwen/qwen3-coder:free do not fall back to the generic 4096-token
-	// defaults in mapOpenRouterModel. The AnyAPI-scoped models.dev entry only
-	// contains a small paid subset and does not cover the free catalog.
-	const enriched = await safeEnrichModelsWithModelsDev(models);
-	return applyHidden(
-		enriched.map((model) => ({
-			...model,
-			_anyapiMetadataVersion: ANYAPI_METADATA_VERSION,
-		})),
-		PROVIDER_ANYAPI,
-	) as AnyApiProviderModel[];
-}
-
-export default async function anyapiProvider(pi: ExtensionAPI) {
-	const apiKey = getAnyapiApiKey();
-	if (!apiKey) {
-		_logger.info("[anyapi] Skipping — ANYAPI_API_KEY not set.");
-		return;
-	}
-
-	const cachedModels = loadProviderCache(PROVIDER_ANYAPI) as
-		| AnyApiProviderModel[]
-		| undefined;
-	const needsMetadataMigration = cachedModels?.some(
-		(model) => model._anyapiMetadataVersion !== ANYAPI_METADATA_VERSION,
-	);
-	const allModels = await loadCachedOrFetchModels(
-		PROVIDER_ANYAPI,
-		() => fetchAnyApiModels(apiKey),
-		// Force one refresh for caches written before context metadata was added;
-		// subsequent warm startups continue using the normal one-hour cache.
-		needsMetadataMigration ? { ttlMs: -1 } : undefined,
-	);
-	if (allModels.length === 0) {
-		_logger.warn("[anyapi] No text models available");
-		return;
-	}
-
-	const freeModels = allModels.filter((model) =>
-		isFreeModel({ ...model, provider: PROVIDER_ANYAPI }, allModels),
-	);
-	const stored = { free: freeModels, all: allModels };
-
-	_logger.info(
-		`[anyapi] Registered ${allModels.length} models (${freeModels.length} free)`,
-	);
-
-	const reRegister = createReRegister(pi, {
-		providerId: PROVIDER_ANYAPI,
-		baseUrl: BASE_URL_ANYAPI,
-		apiKey,
-	});
-
-	registerWithGlobalToggle(PROVIDER_ANYAPI, stored, reRegister, true);
-
-	setupProvider(
-		pi,
-		{
-			providerId: PROVIDER_ANYAPI,
-			initialShowPaid: getAnyapiShowPaid(),
-			hasKey: true,
-			tosUrl: "https://anyapi.ai/terms-of-service",
-			reRegister: (models, current) => {
-				if (current) {
-					stored.free = current.free;
-					stored.all = current.all;
-				}
-				reRegister(models);
-			},
-		},
-		stored,
-	);
-
-	const showPaid = getAnyapiShowPaid();
-	reRegister(showPaid ? stored.all : stored.free);
-}
+/**
+ * AnyAPI provider extension.
+ *
+ * AnyAPI is an OpenAI-compatible gateway with a free plan and a catalog of
+ * explicitly free models. It exposes the catalog at /v1/models and routes
+ * chat requests through /v1/chat/completions.
+ *
+ * Setup:
+ *   ANYAPI_API_KEY=...
+ *   # or add anyapi_api_key to ~/.pi/free.json
+ */
+
+import type {
+	ExtensionAPI,
+	ProviderModelConfig,
+} from "@earendil-works/pi-coding-agent";
+import {
+	applyHidden,
+	getAnyapiApiKey,
+	getAnyapiShowPaid,
+} from "../../config.ts";
+import {
+	BASE_URL_ANYAPI,
+	DEFAULT_FETCH_TIMEOUT_MS,
+	PROVIDER_ANYAPI,
+} from "../../constants.ts";
+import { createLogger } from "../../lib/logger.ts";
+import { loadProviderCache } from "../../lib/provider-cache.ts";
+import { isFreeModel, registerWithGlobalToggle } from "../../lib/registry.ts";
+import { safeEnrichModelsWithModelsDev } from "../../lib/model-metadata.ts";
+import { fetchWithRetry, mapOpenRouterModel } from "../../lib/util.ts";
+import {
+	createReRegister,
+	loadCachedOrFetchModels,
+	setupProvider,
+} from "../../provider-helper.ts";
+
+const _logger = createLogger("anyapi");
+
+interface AnyApiModel {
+	id: string;
+	name?: string;
+	context_length?: number;
+	max_completion_tokens?: number | null;
+	top_provider?: {
+		context_length?: number | null;
+		max_completion_tokens?: number | null;
+	};
+	pricing?: {
+		prompt?: string | number | null;
+		completion?: string | number | null;
+		input_cache_read?: string | number | null;
+		input_cache_write?: string | number | null;
+	};
+	architecture?: {
+		input_modalities?: string[] | null;
+		output_modalities?: string[] | null;
+	};
+	supported_parameters?: string[] | null;
+	tags?: string[];
+	isFree?: boolean;
+}
+
+const ANYAPI_METADATA_VERSION = 1;
+
+type AnyApiProviderModel = ProviderModelConfig & {
+	_pricingKnown?: boolean;
+	_freeKnown?: boolean;
+	_isFree?: boolean;
+	_anyapiMetadataVersion?: number;
+};
+
+function hasPricing(model: AnyApiModel): boolean {
+	return (
+		(model.pricing?.prompt !== null && model.pricing?.prompt !== undefined) ||
+		(model.pricing?.completion !== null &&
+			model.pricing?.completion !== undefined) ||
+		(model.pricing?.input_cache_read !== null &&
+			model.pricing?.input_cache_read !== undefined) ||
+		(model.pricing?.input_cache_write !== null &&
+			model.pricing?.input_cache_write !== undefined)
+	);
+}
+
+function normalizePrice(
+	value: string | number | null | undefined,
+): string | null {
+	return value === null || value === undefined ? null : String(value);
+}
+
+/**
+ * Detect AnyAPI's explicitly free model labels without treating every model
+ * with omitted pricing as free. The API may also expose an authoritative flag
+ * or zero pricing, both of which are handled here.
+ */
+export function isAnyApiFreeModel(model: AnyApiModel): boolean {
+	if (typeof model.isFree === "boolean") return model.isFree;
+
+	const label = `${model.id} ${model.name ?? ""}`.toLowerCase();
+	if (/\bfree\b/.test(label)) return true;
+
+	if (!hasPricing(model)) return false;
+	const input = Number(model.pricing?.prompt);
+	const output = Number(model.pricing?.completion);
+	return input === 0 && output === 0;
+}
+
+export function mapAnyApiModel(model: AnyApiModel): AnyApiProviderModel {
+	const name = model.name ?? model.id;
+	const pricingKnown = hasPricing(model);
+	const freeKnown =
+		typeof model.isFree === "boolean" ||
+		/\bfree\b/i.test(`${model.id} ${name}`) ||
+		(pricingKnown && isAnyApiFreeModel(model));
+
+	const tags = model.tags ?? [];
+	const supportsReasoning =
+		tags.includes("reasoning") || tags.includes("chat_completions:reasoning");
+	const supportsVision =
+		tags.includes("vision") || tags.includes("chat_completions:vision");
+
+	const mapped = mapOpenRouterModel({
+		...model,
+		name,
+		supported_parameters:
+			model.supported_parameters ?? (supportsReasoning ? ["reasoning"] : []),
+		architecture: model.architecture ?? {
+			input_modalities: supportsVision ? ["text", "image"] : ["text"],
+			output_modalities: ["text"],
+		},
+		pricing: model.pricing
+			? {
+					prompt: normalizePrice(model.pricing.prompt),
+					completion: normalizePrice(model.pricing.completion),
+					input_cache_read: normalizePrice(model.pricing.input_cache_read),
+					input_cache_write: normalizePrice(model.pricing.input_cache_write),
+				}
+			: undefined,
+	});
+
+	return {
+		...mapped,
+		_pricingKnown: pricingKnown,
+		...(freeKnown && {
+			_freeKnown: true,
+			_isFree: isAnyApiFreeModel(model),
+		}),
+	};
+}
+
+function isTextModel(model: AnyApiModel): boolean {
+	const outputModalities = model.architecture?.output_modalities ?? [];
+	if (outputModalities.length > 0 && !outputModalities.includes("text")) {
+		return false;
+	}
+
+	const tags = model.tags;
+	if (!tags || tags.length === 0) return true;
+	return tags.some((tag) => tag.startsWith("chat_completions:"));
+}
+
+async function fetchAnyApiModels(
+	apiKey: string,
+): Promise<AnyApiProviderModel[]> {
+	const response = await fetchWithRetry(
+		`${BASE_URL_ANYAPI}/models`,
+		{
+			headers: {
+				Authorization: `Bearer ${apiKey}`,
+				Accept: "application/json",
+				"Content-Type": "application/json",
+			},
+		},
+		3,
+		1000,
+		DEFAULT_FETCH_TIMEOUT_MS,
+	);
+
+	if (!response.ok) {
+		throw new Error(
+			`AnyAPI API error: ${response.status} ${response.statusText}`,
+		);
+	}
+
+	const json = (await response.json()) as { data?: AnyApiModel[] };
+	const models = (json.data ?? []).flatMap((model) => {
+		if (!model.id || !isTextModel(model)) return [];
+		return [mapAnyApiModel(model)];
+	});
+
+	_logger.info(`[anyapi] Fetched ${models.length} text models`);
+
+	// AnyAPI's /models response omits context and output limits for its
+	// catalog. Use the global models.dev catalog so canonical model IDs such
+	// as qwen/qwen3-coder:free do not fall back to the generic 4096-token
+	// defaults in mapOpenRouterModel. The AnyAPI-scoped models.dev entry only
+	// contains a small paid subset and does not cover the free catalog.
+	const enriched = await safeEnrichModelsWithModelsDev(models);
+	return applyHidden(
+		enriched.map((model) => ({
+			...model,
+			_anyapiMetadataVersion: ANYAPI_METADATA_VERSION,
+		})),
+		PROVIDER_ANYAPI,
+	) as AnyApiProviderModel[];
+}
+
+export default async function anyapiProvider(pi: ExtensionAPI) {
+	const apiKey = getAnyapiApiKey();
+	if (!apiKey) {
+		_logger.info("[anyapi] Skipping — ANYAPI_API_KEY not set.");
+		return;
+	}
+
+	const cachedModels = loadProviderCache(PROVIDER_ANYAPI) as
+		| AnyApiProviderModel[]
+		| undefined;
+	const needsMetadataMigration = cachedModels?.some(
+		(model) => model._anyapiMetadataVersion !== ANYAPI_METADATA_VERSION,
+	);
+	const allModels = await loadCachedOrFetchModels(
+		PROVIDER_ANYAPI,
+		() => fetchAnyApiModels(apiKey),
+		// Force one refresh for caches written before context metadata was added;
+		// subsequent warm startups continue using the normal one-hour cache.
+		needsMetadataMigration ? { ttlMs: -1 } : undefined,
+	);
+	if (allModels.length === 0) {
+		_logger.warn("[anyapi] No text models available");
+		return;
+	}
+
+	const freeModels = allModels.filter((model) =>
+		isFreeModel({ ...model, provider: PROVIDER_ANYAPI }, allModels),
+	);
+	const stored = { free: freeModels, all: allModels };
+
+	_logger.info(
+		`[anyapi] Registered ${allModels.length} models (${freeModels.length} free)`,
+	);
+
+	const reRegister = createReRegister(pi, {
+		providerId: PROVIDER_ANYAPI,
+		baseUrl: BASE_URL_ANYAPI,
+		apiKey,
+	});
+
+	registerWithGlobalToggle(PROVIDER_ANYAPI, stored, reRegister, true);
+
+	setupProvider(
+		pi,
+		{
+			providerId: PROVIDER_ANYAPI,
+			initialShowPaid: getAnyapiShowPaid(),
+			hasKey: true,
+			tosUrl: "https://anyapi.ai/terms-of-service",
+			reRegister: (models, current) => {
+				if (current) {
+					stored.free = current.free;
+					stored.all = current.all;
+				}
+				reRegister(models);
+			},
+		},
+		stored,
+	);
+
+	const showPaid = getAnyapiShowPaid();
+	reRegister(showPaid ? stored.all : stored.free);
+}