@diegopetrucci/pi-oracle 0.1.15 → 0.1.19

package/.pi-fleet-tested-version

@@ -1 +1 @@
-0.79.10
+0.80.6

package/index.ts

@@ -7,7 +7,7 @@ import { getAgentDir, getMarkdownTheme, type ExtensionAPI, type ExtensionContext
 import { Container, Markdown, Spacer, Text } from "@earendil-works/pi-tui";
 import { Type } from "typebox";
 
-type ThinkingLevel = "off" | "minimal" | "low" | "medium" | "high" | "xhigh";
+type ThinkingLevel = "off" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max";
 type ThinkingLevelMap = Partial<Record<ThinkingLevel, unknown | null>>;
 
 type PiModel = {
@@ -67,7 +67,7 @@ interface OraclePreferences {
 const READ_ONLY_TOOLS = ["read", "grep", "find", "ls"];
 const READ_ONLY_PLUS_BASH_TOOLS = [...READ_ONLY_TOOLS, "bash"];
 const DEFAULT_THINKING_LEVEL: ThinkingLevel = "xhigh";
-const THINKING_LEVELS = ["off", "minimal", "low", "medium", "high", "xhigh"] as const;
+const THINKING_LEVELS = ["off", "minimal", "low", "medium", "high", "xhigh", "max"] as const;
 const COLLAPSED_LINE_LIMIT = 8;
 const ORACLE_STATUS_ID = "oracle";
 const ORACLE_WIDGET_ID = "oracle";
@@ -82,6 +82,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
 		"claude-opus-4-5",
 		"claude-opus-4-1",
 		"claude-opus-4",
+		"claude-sonnet-5",
 		"claude-sonnet-4-6",
 		"claude-sonnet-4-5",
 		"claude-sonnet-4",
@@ -95,37 +96,32 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
 	anthropic: [
 		"claude-fable-5",
 		"claude-opus-4-8",
-		"claude-opus-4.8",
 		"claude-opus-4-7",
-		"claude-opus-4.7",
 		"claude-opus-4-6",
-		"claude-opus-4.6",
 		"claude-opus-4-5",
-		"claude-opus-4.5",
 		"claude-opus-4-1",
-		"claude-opus-4.1",
-		"claude-opus-4-0",
-		"claude-opus-4",
+		"claude-sonnet-5",
 		"claude-sonnet-4-6",
-		"claude-sonnet-4.6",
 		"claude-sonnet-4-5",
-		"claude-sonnet-4.5",
-		"claude-sonnet-4-0",
-		"claude-sonnet-4",
-		"claude-3-7-sonnet",
 	],
 	"ant-ling": ["Ling-2.6-1T", "Ling-2.6-flash"],
 	"azure-openai-responses": [
+		"gpt-5.6-sol",
+		"gpt-5.6-terra",
+		"gpt-5.6-luna",
 		"gpt-5.5-pro",
 		"gpt-5.5",
 		"gpt-5.4-pro",
 		"gpt-5.4",
 		"gpt-5.3-codex",
 		"gpt-5-pro",
+		"gpt-5-chat-latest",
 		"gpt-5.2-pro",
 		"gpt-5.2",
 		"gpt-5.2-codex",
 		"gpt-5.1-codex-max",
+		"gpt-5.1-codex",
+		"gpt-5.1",
 		"gpt-5.1-chat-latest",
 		"o3-pro",
 		"o3-deep-research",
@@ -133,7 +129,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
 		"gpt-5.4-mini",
 		"gpt-5-mini",
 	],
-	cerebras: ["gpt-oss-120b", "zai-glm-4.7", "llama3.1-8b"],
+	cerebras: ["gpt-oss-120b", "zai-glm-4.7", "gemma-4-31b"],
 	"cloudflare-ai-gateway": [
 		"claude-fable-5",
 		"claude-opus-4-7",
@@ -151,8 +147,8 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
 		"@cf/zai-org/glm-5.2",
 		"@cf/moonshotai/kimi-k2.6",
 		"@cf/nvidia/nemotron-3-120b-a12b",
-		"@cf/moonshotai/kimi-k2.5",
 		"@cf/openai/gpt-oss-120b",
+		"@cf/qwen/qwen3-30b-a3b-fp8",
 		"@cf/zai-org/glm-4.7-flash",
 	],
 	deepseek: ["deepseek-v4-pro", "deepseek-v4-flash"],
@@ -161,13 +157,17 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
 		"accounts/fireworks/models/kimi-k2p7-code",
 		"accounts/fireworks/routers/kimi-k2p7-code-fast",
 		"accounts/fireworks/models/glm-5p2",
+		"accounts/fireworks/routers/glm-5p2-fast",
 		"accounts/fireworks/models/minimax-m3",
 		"accounts/fireworks/models/glm-5p1",
+		"accounts/fireworks/routers/glm-5p1-fast",
 		"accounts/fireworks/models/kimi-k2p6",
+		"accounts/fireworks/routers/kimi-k2p6-turbo",
+		"accounts/fireworks/routers/kimi-k2p6-fast",
 		"accounts/fireworks/models/minimax-m2p7",
 		"accounts/fireworks/models/qwen3p7-plus",
-		"accounts/fireworks/models/qwen3p6-plus",
 		"accounts/fireworks/models/gpt-oss-120b",
+		"accounts/fireworks/models/gpt-oss-20b",
 	],
 	"github-copilot": [
 		"claude-fable-5",
@@ -178,40 +178,46 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
 		"gpt-5.5",
 		"gpt-5.4",
 		"gpt-5.3-codex",
+		"gpt-5.2-codex",
 		"gpt-5.2",
-		"gpt-5.1-codex-max",
-		"gpt-5.1",
-		"gpt-5",
 		"gemini-3.1-pro-preview",
-		"gemini-3-pro-preview",
+		"gemini-3.5-flash",
+		"gemini-3-flash-preview",
+		"claude-sonnet-5",
 		"claude-sonnet-4.6",
 		"claude-sonnet-4.5",
 		"gemini-2.5-pro",
+		"gpt-5-mini",
+		"kimi-k2.7-code",
 	],
 	google: [
 		"gemini-3.1-pro-preview-customtools",
 		"gemini-3.1-pro-preview",
 		"gemini-3-pro-preview",
-		"gemini-2.5-pro-preview",
+		"gemini-3.5-flash",
 		"gemini-2.5-pro",
-		"gemini-2.5-flash-preview",
-		"gemini-2.5-flash-lite-preview",
+		"gemini-2.5-flash",
+		"gemini-3.1-flash-lite-preview",
+		"gemini-3.1-flash-lite",
 		"gemini-2.5-flash-lite",
+		"gemini-2.0-flash",
 	],
 	"google-vertex": [
 		"gemini-3.1-pro-preview-customtools",
 		"gemini-3.1-pro-preview",
-		"gemini-3-pro-preview",
+		"gemini-3.5-flash",
+		"gemini-3-flash-preview",
 		"gemini-2.5-pro",
+		"gemini-2.5-flash",
+		"gemini-3.1-flash-lite",
 		"gemini-2.5-flash-lite",
-		"gemini-2.0-flash-lite",
 	],
 	groq: [
 		"openai/gpt-oss-120b",
-		"groq/compound-mini",
 		"qwen/qwen3-32b",
-		"moonshotai/kimi-k2-instruct",
-		"meta-llama/llama-4-scout",
+		"meta-llama/llama-4-scout-17b-16e-instruct",
+		"llama-3.3-70b-versatile",
+		"openai/gpt-oss-20b",
 	],
 	huggingface: [
 		"zai-org/GLM-5.1",
@@ -241,16 +247,22 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
 		"devstral-2512",
 	],
 	openai: [
+		"gpt-5.6-sol",
+		"gpt-5.6-terra",
+		"gpt-5.6-luna",
 		"gpt-5.5-pro",
 		"gpt-5.5",
 		"gpt-5.4-pro",
 		"gpt-5.4",
 		"gpt-5.3-codex",
 		"gpt-5-pro",
+		"gpt-5-chat-latest",
 		"gpt-5.2-pro",
 		"gpt-5.2",
 		"gpt-5.2-codex",
 		"gpt-5.1-codex-max",
+		"gpt-5.1-codex",
+		"gpt-5.1",
 		"gpt-5.1-chat-latest",
 		"o3-pro",
 		"o3-deep-research",
@@ -259,14 +271,13 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
 		"gpt-5-mini",
 	],
 	"openai-codex": [
+		"gpt-5.6-sol",
+		"gpt-5.6-terra",
+		"gpt-5.6-luna",
 		"gpt-5.5",
 		"gpt-5.4",
-		"gpt-5.3-codex",
-		"gpt-5.2",
-		"gpt-5.1-codex-max",
 		"gpt-5.4-mini",
-		"gpt-5.1-codex-mini",
-		"big-pickle",
+		"gpt-5.3-codex-spark",
 	],
 	opencode: [
 		"claude-fable-5",
@@ -281,15 +292,19 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
 		"gpt-5.3-codex",
 		"gpt-5.2-codex",
 		"gpt-5.1-codex",
+		"grok-4.5",
+		"gemini-3.5-flash",
 		"gemini-3.1-pro",
+		"glm-5.2",
 		"glm-5.1",
 		"glm-5",
+		"kimi-k2.7-code",
 		"kimi-k2.6",
 		"kimi-k2.5",
 		"qwen3.6-plus",
 		"qwen3.5-plus",
+		"minimax-m3",
 		"minimax-m2.7",
-		"minimax-m2.5-free",
 	],
 	"opencode-go": [
 		"deepseek-v4-pro",
@@ -299,12 +314,11 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
 		"qwen3.7-plus",
 		"qwen3.6-plus",
 		"mimo-v2.5-pro",
-		"mimo-v2-pro",
+		"mimo-v2.5",
 		"minimax-m3",
 		"minimax-m2.7",
 		"kimi-k2.7-code",
 		"kimi-k2.6",
-		"kimi-k2.5",
 	],
 	openrouter: [
 		"anthropic/claude-fable-5",
@@ -312,20 +326,30 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
 		"anthropic/claude-opus-4.8",
 		"anthropic/claude-opus-4.8-fast",
 		"anthropic/claude-opus-4.7",
-		"anthropic/claude-opus-4.6-fast",
+		"anthropic/claude-opus-4.7-fast",
 		"anthropic/claude-opus-4.6",
 		"anthropic/claude-opus-4.5",
-		"anthropic/claude-opus-4",
+		"anthropic/claude-opus-4.1",
+		"openai/gpt-5.6-sol-pro",
+		"openai/gpt-5.6-sol",
+		"openai/gpt-5.6-terra-pro",
+		"openai/gpt-5.6-terra",
+		"openai/gpt-5.6-luna-pro",
+		"openai/gpt-5.6-luna",
 		"openai/gpt-5.5-pro",
 		"openai/gpt-5.5",
-		"openai/gpt-5.4-pro",
-		"openai/gpt-5.4",
+		"~openai/gpt-latest",
 		"google/gemini-3.1-pro-preview-customtools",
 		"google/gemini-3.1-pro-preview",
+		"~google/gemini-pro-latest",
+		"google/gemini-3.5-flash",
 		"google/gemini-2.5-pro",
 		"moonshotai/kimi-k2.7-code",
+		"~moonshotai/kimi-latest",
 		"moonshotai/kimi-k2.6",
 		"moonshotai/kimi-k2-thinking",
+		"x-ai/grok-4.5",
+		"~x-ai/grok-latest",
 		"deepseek/deepseek-v4-pro",
 		"deepseek/deepseek-r1",
 		"deepseek/deepseek-v3.2",
@@ -337,6 +361,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
 	],
 	together: [
 		"deepseek-ai/DeepSeek-V4-Pro",
+		"zai-org/GLM-5.2",
 		"zai-org/GLM-5.1",
 		"moonshotai/Kimi-K2.7-Code",
 		"moonshotai/Kimi-K2.6",
@@ -345,12 +370,10 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
 		"MiniMaxAI/MiniMax-M3",
 		"MiniMaxAI/MiniMax-M2.7",
 		"Qwen/Qwen3.5-397B-A17B",
-		"Qwen/Qwen3-Coder-Next-FP8",
-		"Qwen/Qwen3-235B-A22B-Instruct-2507-tput",
 		"openai/gpt-oss-120b",
-		"moonshotai/Kimi-K2.5",
-		"deepseek-ai/DeepSeek-V3-1",
-		"MiniMaxAI/MiniMax-M2.5",
+		"openai/gpt-oss-20b",
+		"nvidia/nemotron-3-ultra-550b-a55b",
+		"google/gemma-4-31B-it",
 	],
 	"vercel-ai-gateway": [
 		"anthropic/claude-fable-5",
@@ -359,11 +382,17 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
 		"anthropic/claude-opus-4.6",
 		"anthropic/claude-opus-4.5",
 		"anthropic/claude-opus-4.1",
+		"anthropic/claude-sonnet-5",
 		"anthropic/claude-sonnet-4.6",
+		"openai/gpt-5.6-sol",
+		"openai/gpt-5.6-terra",
+		"openai/gpt-5.6-luna",
 		"openai/gpt-5.5-pro",
 		"openai/gpt-5.5",
 		"openai/gpt-5.4-pro",
 		"openai/gpt-5.4",
+		"openai/gpt-5.3-codex",
+		"openai/gpt-5.2-codex",
 		"openai/gpt-5.1-codex",
 		"openai/gpt-5-codex",
 		"moonshotai/kimi-k2.7-code",
@@ -374,52 +403,37 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
 		"deepseek/deepseek-v3.2-thinking",
 		"alibaba/qwen3.7-max",
 		"alibaba/qwen3.7-plus",
+		"alibaba/qwen3.6-plus",
 		"alibaba/qwen3.5-plus",
 		"alibaba/qwen3-max-thinking",
 		"google/gemini-3.1-pro-preview",
+		"google/gemini-3.5-flash",
 		"google/gemini-3-flash",
-		"xai/grok-4.3",
+		"xai/grok-4.5",
+		"xai/grok-4.20-reasoning",
 		"minimax/minimax-m3",
 		"zai/glm-5.2",
 		"zai/glm-5.1",
 	],
 	xai: [
-		"grok-4.3",
+		"grok-4.5",
 		"grok-4.20-0309-reasoning",
-		"grok-4-1-fast",
-		"grok-4-fast",
-		"grok-4",
-		"grok-3-mini-latest",
-		"grok-3-mini-fast",
-		"grok-3-latest",
+		"grok-4.20-0309-non-reasoning",
+		"grok-4.3",
+		"grok-build-0.1",
+		"grok-code-fast-1",
+		"grok-3-fast",
+		"grok-3",
 	],
 	xiaomi: ["mimo-v2.5-pro", "mimo-v2.5", "mimo-v2-pro", "mimo-v2-omni", "mimo-v2-flash"],
-	"xiaomi-token-plan-ams": [
-		"mimo-v2.5-pro",
-		"mimo-v2.5",
-		"mimo-v2-pro",
-		"mimo-v2-omni",
-		"mimo-v2-flash",
-	],
-	"xiaomi-token-plan-cn": [
-		"mimo-v2.5-pro",
-		"mimo-v2.5",
-		"mimo-v2-pro",
-		"mimo-v2-omni",
-		"mimo-v2-flash",
-	],
-	"xiaomi-token-plan-sgp": [
-		"mimo-v2.5-pro",
-		"mimo-v2.5",
-		"mimo-v2-pro",
-		"mimo-v2-omni",
-		"mimo-v2-flash",
-	],
+	"xiaomi-token-plan-ams": ["mimo-v2.5-pro", "mimo-v2.5", "mimo-v2-pro"],
+	"xiaomi-token-plan-cn": ["mimo-v2.5-pro", "mimo-v2.5", "mimo-v2-pro"],
+	"xiaomi-token-plan-sgp": ["mimo-v2.5-pro", "mimo-v2.5", "mimo-v2-pro"],
 	nvidia: [
 		"nvidia/nemotron-3-ultra-550b-a55b",
 		"nvidia/nemotron-3-super-120b-a12b",
 		"moonshotai/kimi-k2.6",
-		"z-ai/glm-5.1",
+		"z-ai/glm-5.2",
 		"qwen/qwen3.5-122b-a10b",
 		"openai/gpt-oss-120b",
 		"nvidia/nemotron-3-nano-omni-30b-a3b-reasoning",
@@ -430,11 +444,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
 		"glm-5.1",
 		"glm-5-turbo",
 		"glm-5v-turbo",
-		"glm-5",
 		"glm-4.7",
-		"glm-4.7-flash",
-		"glm-4.6v",
-		"glm-4.5v",
 		"glm-4.5-air",
 	],
 	"zai-coding-cn": [
@@ -516,7 +526,7 @@ function normalizeModelPreference(value: unknown): string | undefined {
 function parseModelPreference(value: unknown): { model?: string; thinkingLevel?: ThinkingLevel } {
 	const model = normalizeModelPreference(value);
 	if (!model) return {};
-	const match = model.match(/^(.*):(off|minimal|low|medium|high|xhigh)$/i);
+	const match = model.match(/^(.*):(off|minimal|low|medium|high|xhigh|max)$/i);
 	if (!match?.[1]) return { model };
 	return { model: match[1], thinkingLevel: parseThinkingLevel(match[2]) };
 }
@@ -612,6 +622,19 @@ function isTransientErrorMessage(message: string): boolean {
 	return TRANSIENT_ERROR_PATTERN.test(message);
 }
 
+// A model the catalog advertises but the active provider/subscription cannot
+// serve surfaces as a not-found/404-style error (legacy snapshots or
+// access-gated tiers). These are NOT transient: the same model keeps failing,
+// so callers should fall back to a different model instead of retrying it.
+const MODEL_AVAILABILITY_ERROR_PATTERN =
+	/\b(?:404|403|not[_ ]?found(?:[_ ]?error)?|model[_ ]?not[_ ]?found(?:[_ ]?error)?|no such model|unknown model|does not exist|is not available|not available|model[_ ]?not[_ ]?available|unsupported model|invalid model|forbidden|access[ _-]?denied|permission[ _-]?denied|not[ _-]?entitled|do(?:es)? not have access)\b/i;
+
+function isModelAvailabilityError(message: string | undefined): boolean {
+	if (!message) return false;
+	if (isTransientErrorMessage(message)) return false;
+	return MODEL_AVAILABILITY_ERROR_PATTERN.test(message);
+}
+
 function formatOracleModelError(stopReason: "error" | "aborted", errorMessage: string | undefined): string {
 	const trimmed = errorMessage?.trim();
 	if (stopReason === "aborted") {
@@ -700,7 +723,7 @@ function getPiInvocation(args: string[]): { command: string; args: string[] } {
 }
 
 function withThinking(modelRef: string, thinkingLevel: ThinkingLevel): string {
-	if (/(?:^|\/)[^:]+:(off|minimal|low|medium|high|xhigh)$/i.test(modelRef)) return modelRef;
+	if (/(?:^|\/)[^:]+:(off|minimal|low|medium|high|xhigh|max)$/i.test(modelRef)) return modelRef;
 	return `${modelRef}:${thinkingLevel}`;
 }
 
@@ -709,8 +732,8 @@ function isThinkingLevelSupported(model: PiModel, level: ThinkingLevel): boolean
 	if (!model.reasoning) return level === "off";
 
 	const map = model.thinkingLevelMap;
-	if (level === "xhigh") {
-		return !!map && Object.prototype.hasOwnProperty.call(map, "xhigh") && map.xhigh != null;
+	if (level === "xhigh" || level === "max") {
+		return !!map && Object.prototype.hasOwnProperty.call(map, level) && map[level] != null;
 	}
 	return map?.[level] !== null;
 }
@@ -731,11 +754,16 @@ function clampThinkingLevel(model: PiModel, requested: ThinkingLevel): ThinkingL
 	return "off";
 }
 
+function isGpt56SolModel(model: PiModel | undefined): boolean {
+	if (!model) return false;
+	return /\bgpt-5\.6-sol\b/i.test(`${model.id} ${model.name ?? ""}`);
+}
+
 function resolveThinkingLevel(
 	model: PiModel | undefined,
 	override: ThinkingLevel | undefined,
 ): { requested: ThinkingLevel; effective: ThinkingLevel; clamped: boolean } {
-	const requested = override ?? (model?.reasoning ? DEFAULT_THINKING_LEVEL : "off");
+	const requested = override ?? (model?.reasoning ? (isGpt56SolModel(model) ? "high" : DEFAULT_THINKING_LEVEL) : "off");
 	const effective = model ? clampThinkingLevel(model, requested) : requested;
 	return { requested, effective, clamped: effective !== requested };
 }
@@ -776,10 +804,48 @@ async function findAvailableModel(
 	return undefined;
 }
 
+function toOracleSelection(
+	model: PiModel,
+	thinkingLevelOverride: ThinkingLevel | undefined,
+	reason: string,
+): OracleSelection {
+	const thinking = resolveThinkingLevel(model, thinkingLevelOverride);
+	return {
+		modelRef: `${model.provider}/${model.id}`,
+		provider: model.provider,
+		modelId: model.id,
+		modelName: model.name,
+		thinkingLevel: thinking.effective,
+		...(thinking.clamped ? { requestedThinkingLevel: thinking.requested, thinkingLevelClamped: true } : {}),
+		autoSelected: true,
+		selectionReason: appendThinkingLevelClampReason(reason, thinking),
+	};
+}
+
+function buildSessionFallbackSelection(
+	ctx: { model?: PiModel },
+	thinkingLevelOverride: ThinkingLevel | undefined,
+): OracleSelection | undefined {
+	const model = ctx.model;
+	if (!model) return undefined;
+	return toOracleSelection(
+		model,
+		thinkingLevelOverride,
+		`Used the current session model ${model.provider}/${model.id} as a final fallback.`,
+	);
+}
+
+function withFallbackReason(selection: OracleSelection, previous: OracleSelection): OracleSelection {
+	return {
+		...selection,
+		selectionReason: `${selection.selectionReason} (Fell back from ${previous.modelRef} after a model-availability error.)`,
+	};
+}
+
 async function selectOracleModel(
 	ctx: { model?: PiModel; modelRegistry: { getAvailable(): PiModel[] | Promise<PiModel[]> } },
 	thinkingLevelOverride?: ThinkingLevel,
-): Promise<{ ok: true; selection: OracleSelection } | { ok: false; error: string }> {
+): Promise<{ ok: true; selection: OracleSelection; ordered: OracleSelection[] } | { ok: false; error: string }> {
 	const available = await ctx.modelRegistry.getAvailable();
 	if (available.length === 0) {
 		return {
@@ -810,27 +876,23 @@ async function selectOracleModel(
 		reason = "No reasoning models were available, so the top-ranked model across all available providers was used.";
 	}
 
+	const sorted = [...candidates].sort((a, b) => rankModel(b) - rankModel(a));
 	const preferred = selectPreferredModel(candidates, providerForPreferences);
-	const winner = preferred ?? [...candidates].sort((a, b) => rankModel(b) - rankModel(a))[0];
-	const thinking = resolveThinkingLevel(winner, thinkingLevelOverride);
-	const selectionReason = appendThinkingLevelClampReason(
-		preferred ? `Selected ${winner.id} via the hardcoded preference list for ${winner.provider}.` : reason,
-		thinking,
+	const orderedModels = preferred ? [preferred, ...sorted.filter((model) => model !== preferred)] : sorted;
+
+	const ordered = orderedModels.map((model, index) =>
+		toOracleSelection(
+			model,
+			thinkingLevelOverride,
+			index === 0
+				? preferred
+					? `Selected ${model.id} via the hardcoded preference list for ${model.provider}.`
+					: reason
+				: `Auto-selected ${model.provider}/${model.id} as a lower-ranked fallback.`,
+		),
 	);
 
-	return {
-		ok: true,
-		selection: {
-			modelRef: `${winner.provider}/${winner.id}`,
-			provider: winner.provider,
-			modelId: winner.id,
-			modelName: winner.name,
-			thinkingLevel: thinking.effective,
-			...(thinking.clamped ? { requestedThinkingLevel: thinking.requested, thinkingLevelClamped: true } : {}),
-			autoSelected: true,
-			selectionReason,
-		},
-	};
+	return { ok: true, selection: ordered[0], ordered };
 }
 
 function updateOracleUi(ctx: ExtensionContext, activeRuns: Map<string, OracleUiRun>): void {
@@ -1061,6 +1123,13 @@ function renderCollapsedText(text: string, lineLimit = COLLAPSED_LINE_LIMIT): st
 	return [...lines.slice(0, lineLimit), `... (${lines.length - lineLimit} more lines)`].join("\n");
 }
 
+export const __test__ = {
+	findAvailableModel,
+	parseModelPreference,
+	resolveThinkingLevel,
+	selectOracleModel,
+};
+
 export default function oracleExtension(pi: ExtensionAPI) {
 	const activeRuns = new Map<string, OracleUiRun>();
 	let preferences: OraclePreferences = {};
@@ -1177,7 +1246,7 @@ export default function oracleExtension(pi: ExtensionAPI) {
 				return;
 			}
 
-			notifyCommand(ctx, "Usage: /oracle status | model <provider/model|auto> | thinking <off|minimal|low|medium|high|xhigh|auto> | clear [all|model|thinking]", "warning");
+			notifyCommand(ctx, "Usage: /oracle status | model <provider/model|auto> | thinking <off|minimal|low|medium|high|xhigh|max|auto> | clear [all|model|thinking]", "warning");
 		},
 	});
 
@@ -1223,13 +1292,17 @@ export default function oracleExtension(pi: ExtensionAPI) {
 		parameters: OracleParams,
 
 		async execute(toolCallId, params, signal, onUpdate, ctx) {
+			const rawTask = (params as { task?: unknown }).task;
+			const task = typeof rawTask === "string" ? rawTask.trim() : "";
+			if (!task) throw new Error("Invalid parameters: expected task to be a non-empty string.");
+
 			const explicitModel = parseModelPreference(params.model);
 			const defaultModel = parseModelPreference(preferences.model);
 			const configuredModel = explicitModel.model ?? defaultModel.model;
 			const configuredThinkingLevel =
 				params.thinkingLevel ?? explicitModel.thinkingLevel ?? preferences.thinkingLevel ?? defaultModel.thinkingLevel;
 			const uiRun: OracleUiRun = {
-				task: params.task,
+				task,
 				includeBash: params.includeBash ?? false,
 				startedAt: Date.now(),
 			};
@@ -1237,7 +1310,16 @@ export default function oracleExtension(pi: ExtensionAPI) {
 			updateOracleUi(ctx, activeRuns);
 
 			try {
-				let selection: OracleSelection;
+				const attempts: OracleSelection[] = [];
+				const seen = new Set<string>();
+				const pushAttempt = (candidate: OracleSelection | undefined) => {
+					if (!candidate) return;
+					const key = `${candidate.provider}/${candidate.modelId}`.toLowerCase();
+					if (seen.has(key)) return;
+					seen.add(key);
+					attempts.push(candidate);
+				};
+
 				if (configuredModel) {
 					const modelRef = configuredModel;
 					const matched = await findAvailableModel(ctx, modelRef);
@@ -1256,7 +1338,7 @@ export default function oracleExtension(pi: ExtensionAPI) {
 						: usedToolOverride
 							? "Used the explicit model override provided in the tool call. The model was not matched against the authenticated model list, so the reasoning level fallback was applied."
 							: "Used the configured default oracle model. The model was not matched against the authenticated model list, so the reasoning level fallback was applied.";
-					selection = {
+					pushAttempt({
 						modelRef: matched ? `${matched.provider}/${matched.id}` : modelRef,
 						provider,
 						modelId,
@@ -1267,7 +1349,7 @@ export default function oracleExtension(pi: ExtensionAPI) {
 							: {}),
 						autoSelected: false,
 						selectionReason,
-					};
+					});
 				} else {
 					const selectionResult = await selectOracleModel(ctx, configuredThinkingLevel);
 					if (!selectionResult.ok) {
@@ -1290,11 +1372,14 @@ export default function oracleExtension(pi: ExtensionAPI) {
 							},
 						};
 					}
-					selection = selectionResult.selection;
+					for (const candidate of selectionResult.ordered) pushAttempt(candidate);
 				}
 
-				uiRun.selection = selection;
-				updateOracleUi(ctx, activeRuns);
+				// Keep the known-good session model as a final fallback: the catalog can
+				// advertise models the active provider/subscription cannot actually serve
+				// (legacy snapshots, access-gated tiers), which fail with a not-found/404
+				// error. Falling back lets the run degrade gracefully instead of hard-failing.
+				pushAttempt(buildSessionFallbackSelection(ctx, configuredThinkingLevel));
 
 				const handleUpdate = (partial: { content: Array<{ type: "text"; text: string }>; details: OracleDetails }) => {
 					uiRun.preview = partial.content[0]?.text ?? uiRun.preview;
@@ -1302,17 +1387,51 @@ export default function oracleExtension(pi: ExtensionAPI) {
 					onUpdate?.(partial);
 				};
 
-				const result = await runOracle(selection, params, signal, handleUpdate, ctx.cwd);
-				if (!result.ok) {
-					return {
-						content: [{ type: "text", text: result.error }],
-						details: result.details,
-					};
+				let lastErrorResult: { ok: false; error: string; details: OracleDetails } | undefined;
+				for (let index = 0; index < attempts.length; index++) {
+					const previous = index > 0 ? attempts[index - 1] : undefined;
+					const attempt = previous ? withFallbackReason(attempts[index], previous) : attempts[index];
+					uiRun.selection = attempt;
+					updateOracleUi(ctx, activeRuns);
+
+					const result = await runOracle(attempt, { ...params, task }, signal, handleUpdate, ctx.cwd);
+					if (result.ok) {
+						return {
+							content: [{ type: "text", text: result.output }],
+							details: result.details,
+						};
+					}
+
+					lastErrorResult = result;
+					const canFallBack = index < attempts.length - 1 && isModelAvailabilityError(result.error);
+					if (!canFallBack) {
+						return {
+							content: [{ type: "text", text: result.error }],
+							details: result.details,
+						};
+					}
 				}
 
 				return {
-					content: [{ type: "text", text: result.output }],
-					details: result.details,
+					content: [
+						{ type: "text", text: lastErrorResult?.error ?? "Oracle could not select an available model." },
+					],
+					details:
+						lastErrorResult?.details ?? {
+							modelRef: "",
+							provider: ctx.model?.provider ?? "unknown",
+							modelId: "",
+							modelName: undefined,
+							thinkingLevel: configuredThinkingLevel ?? DEFAULT_THINKING_LEVEL,
+							autoSelected: true,
+							selectionReason: "No authenticated models were available to run the oracle.",
+							includeBash: params.includeBash ?? false,
+							usage: createEmptyUsage(),
+							stderr: "",
+							exitCode: 1,
+							durationMs: 0,
+							cwd: params.cwd ?? ctx.cwd,
+						},
 				};
 			} finally {
 				activeRuns.delete(toolCallId);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@diegopetrucci/pi-oracle",
-  "version": "0.1.15",
+  "version": "0.1.19",
   "description": "An Amp-style oracle extension for pi that consults the strongest reasoning model on your current provider.",
   "keywords": [
     "pi-package",