@agimon-ai/doompi-cache 0.0.1-alpha.81 → 0.0.1-alpha.83

package/dist/{pi-cache-optimizer-VJG1vKKU.cjs → pi-cache-optimizer-DSED41mZ.cjs}

@@ -4,7 +4,7 @@ let node_fs = require("node:fs");
 let node_os = require("node:os");
 let node_path = require("node:path");
 let _earendil_works_pi_coding_agent = require("@earendil-works/pi-coding-agent");
-//#region ../../../node_modules/.pnpm/pi-cache-optimizer@2.8.11_@earendil-works+pi-coding-agent@1.0.2_@aws-sdk+credential-pro_d92cfe1c8329e9bda023a760e3edd654/node_modules/pi-cache-optimizer/index.ts
+//#region ../../../node_modules/.pnpm/pi-cache-optimizer@2.8.16_@earendil-works+pi-coding-agent@1.0.2_@aws-sdk+credential-pro_336a3137387524f540927fd810e3b800/node_modules/pi-cache-optimizer/index.ts
 const PI_CACHE_RETENTION_ENV = "PI_CACHE_RETENTION";
 const LONG_CACHE_RETENTION_VALUE = "long";
 const PI_CACHE_RETENTION_BASELINE_SYMBOL = Symbol.for("pi.cache.optimizer.retention-baseline.v1");
@@ -91,6 +91,7 @@ const NO_OPENAI_CACHE_KEY_ENV = "PI_CACHE_OPTIMIZER_NO_OPENAI_CACHE_KEY";
 const OPENAI_PROMPT_CACHE_KEY_MAX_LENGTH = 64;
 const NO_SKILL_COMPRESSION_ENV = "PI_CACHE_OPTIMIZER_NO_SKILL_COMPRESSION";
 const NO_PROMPT_REWRITE_ENV = "PI_CACHE_OPTIMIZER_NO_PROMPT_REWRITE";
+const VIRTUAL_REWRITE_ENV = "PI_CACHE_OPTIMIZER_VIRTUAL_REWRITE";
 const TOOL_ORDER_ENV = "PI_CACHE_OPTIMIZER_TOOL_ORDER";
 const FOOTER_MODE_ENV = "PI_CACHE_OPTIMIZER_FOOTER_MODE";
 const RECEIPT_COMPAT_KEYS = /* @__PURE__ */ new Set([
@@ -104,6 +105,7 @@ const RECEIPT_COMPAT_KEYS = /* @__PURE__ */ new Set([
 ]);
 const PI_ROUTING_REGISTRY_SYMBOL = Symbol.for("pi.routing.registry.v1");
 const PI_CACHE_HINTS_SYMBOL = Symbol.for("pi.cache.hints.v1");
+const PI_VIRTUAL_MODEL_API = "pi-virtual";
 const PI_CACHE_HINTS_OWNER_SYMBOL = Symbol.for("pi.cache.optimizer.hints-owner.v1");
 const ANTHROPIC_TTL_FALLBACK_SYMBOL = Symbol.for("pi.cache.optimizer.anthropic-ttl-fallback.v1");
 const REASONING_PROTOCOL_FALLBACK_SYMBOL = Symbol.for("pi.cache.optimizer.reasoning-protocol-fallback.v1");
@@ -437,7 +439,7 @@ function formatSkillsForPromptCompressed(skills) {
 *     than mangle.
 */
 function compressSkillsInSystemPrompt(prompt, opts) {
-	if (isEnabledEnv(process.env[NO_SKILL_COMPRESSION_ENV])) return prompt;
+	if (!featureEnabled("skillCompression", NO_SKILL_COMPRESSION_ENV, true)) return prompt;
 	if (!opts.skills || opts.skills.length === 0) return prompt;
 	if (opts.skills.filter((skill) => !skill.disableModelInvocation).length < SKILL_COMPRESSION_MIN_COUNT) return prompt;
 	const verbose = formatSkillsForPrompt(opts.skills);
@@ -692,7 +694,7 @@ function parseRouteSnapshot(value, fallbackVirtualProvider, fallbackVirtualModel
 	};
 }
 function resolveActiveRouteSnapshot(model, ctx) {
-	if (!model) return void 0;
+	if (!model || isNativeVirtualModel(model)) return void 0;
 	const hint = ctx ? { sessionIdHash: sessionHashFromContext(ctx) } : void 0;
 	const adapter = getRoutingRegistry()?.getRouter(model.provider);
 	if (adapter) try {
@@ -713,6 +715,31 @@ function resolveActiveRouteSnapshot(model, ctx) {
 		return;
 	}
 }
+function resolveNativeVirtualCandidateModels(model, ctx) {
+	if (!isNativeVirtualModel(model)) return void 0;
+	const adapter = model ? getRoutingRegistry()?.getRouter(model.provider) : void 0;
+	if (!adapter?.resolveCandidateRoutes) return void 0;
+	try {
+		const rawSnapshots = adapter.resolveCandidateRoutes(model.id);
+		if (!Array.isArray(rawSnapshots) || rawSnapshots.length === 0) return void 0;
+		const candidates = [];
+		for (const rawSnapshot of rawSnapshots) {
+			const snapshot = parseRouteSnapshot(rawSnapshot, model.provider, model.id);
+			if (!snapshot) return void 0;
+			const candidate = findModelInRegistry(ctx?.modelRegistry, snapshot.provider, snapshot.modelId) ?? routeSnapshotToPiModel(snapshot, model);
+			if (!isNonEmptyString(candidate.api)) return void 0;
+			candidates.push(candidate);
+		}
+		return candidates;
+	} catch {
+		return;
+	}
+}
+function canRewriteNativeVirtualPrompt(model, ctx) {
+	if (!isNativeVirtualModel(model) || !featureEnabled("virtualRewrite", VIRTUAL_REWRITE_ENV, false)) return false;
+	const candidates = resolveNativeVirtualCandidateModels(model, ctx);
+	return !!candidates?.length && candidates.every((candidate) => candidate.api === "openai-completions" || candidate.api === "anthropic-messages") && candidates.every((candidate) => !isResponsesPromptRewriteBypassApi(candidate.api));
+}
 function routeSnapshotToPiModel(snapshot, fallback) {
 	const sameIdentity = fallback?.provider === snapshot.provider && fallback?.id === snapshot.modelId;
 	return {
@@ -769,13 +796,151 @@ function applyConfiguredTransportToModel(model, config) {
 	};
 }
 function resolveRouteModel(model, ctx) {
+	const nativeVirtualRoute = resolveNativeVirtualRouteModel(model, ctx);
+	if (nativeVirtualRoute) return nativeVirtualRoute;
 	const snapshot = resolveActiveRouteSnapshot(model, ctx);
 	if (!snapshot) return void 0;
 	return applyConfiguredTransportToModel(findModelInRegistry(ctx?.modelRegistry, snapshot.provider, snapshot.modelId) ?? routeSnapshotToPiModel(snapshot, model), readEffectiveCompatConfig());
 }
 function isVirtualRoutingModel(model, ctx) {
 	if (!model) return false;
-	return isRouterModel(model) || !!getRoutingRegistry()?.getRouter(model.provider) || !!resolveActiveRouteSnapshot(model, ctx);
+	return isNativeVirtualModel(model) || isRouterModel(model) || !!getRoutingRegistry()?.getRouter(model.provider) || !!resolveActiveRouteSnapshot(model, ctx);
+}
+function isNativeVirtualModel(model) {
+	return model?.api === PI_VIRTUAL_MODEL_API;
+}
+function readSessionBranch(ctx) {
+	try {
+		const manager = ctx?.sessionManager;
+		if (typeof manager?.getBranch !== "function") return [];
+		const branch = manager.getBranch();
+		return Array.isArray(branch) ? branch : [];
+	} catch {
+		return [];
+	}
+}
+/**
+* Physical models that answered on the current session branch, newest first:
+* the latest successful response (Pi's `request.previous`) and the latest
+* response of any outcome (a failed attempt that a retry may reuse). The
+* catalog id in `message.model` is preferred over the echoed `responseModel`
+* because it is the id Pi dispatched and the registry knows.
+*/
+function findNativeVirtualDispatches(ctx) {
+	const branch = readSessionBranch(ctx);
+	let latestSuccessful;
+	let latestAny;
+	for (let index = branch.length - 1; index >= 0 && !latestSuccessful; index--) {
+		const entry = asRecord(branch[index]);
+		if (entry?.type !== "message") continue;
+		const message = getAssistantRecord(entry.message);
+		if (!message || isNativeVirtualModel(message)) continue;
+		const dispatch = nativeVirtualDispatchFromMessage(message);
+		if (!dispatch) continue;
+		latestAny ??= dispatch;
+		if (message.stopReason !== "error" && message.stopReason !== "aborted") latestSuccessful = dispatch;
+	}
+	return {
+		latestSuccessful,
+		latestAny
+	};
+}
+function nativeVirtualDispatchFromMessage(message) {
+	const record = getAssistantRecord(message);
+	if (!record || isNativeVirtualModel(record)) return void 0;
+	const provider = firstNonEmptyString(record.provider);
+	const id = firstNonEmptyString(record.model, record.responseModel);
+	return provider && id ? {
+		provider,
+		id,
+		api: firstNonEmptyString(record.api) ?? ""
+	} : void 0;
+}
+function nativeVirtualDispatchToModel(dispatch, ctx) {
+	const registered = findModelInRegistry(ctx?.modelRegistry, dispatch.provider, dispatch.id);
+	return applyConfiguredTransportToModel(registered && !isNativeVirtualModel(registered) ? registered : {
+		provider: dispatch.provider,
+		id: dispatch.id,
+		name: dispatch.id,
+		api: dispatch.api,
+		baseUrl: "",
+		reasoning: false,
+		input: ["text"],
+		cost: {
+			input: 0,
+			output: 0,
+			cacheRead: 0,
+			cacheWrite: 0
+		},
+		contextWindow: 0,
+		maxTokens: 0,
+		[ROUTED_FALLBACK_MODEL_SYMBOL]: true
+	}, readEffectiveCompatConfig());
+}
+/**
+* Physical model behind a native virtual selection for pre-request UX (footer,
+* doctor, compat, stats, reset, fix): the model that answered last on this
+* session branch, matching Pi's own routed-model display and context limits.
+*/
+function resolveNativeVirtualRouteModel(model, ctx) {
+	if (!isNativeVirtualModel(model)) return void 0;
+	const { latestSuccessful, latestAny } = findNativeVirtualDispatches(ctx);
+	const dispatch = latestSuccessful ?? latestAny;
+	return dispatch ? nativeVirtualDispatchToModel(dispatch, ctx) : void 0;
+}
+function getProviderPayloadModelId(payload) {
+	const record = asRecord(payload);
+	return firstNonEmptyString(record?.model, record?.modelId);
+}
+function physicalModelsWithId(candidates, id) {
+	return (candidates ?? []).filter((candidate) => candidate.id === id && !isNativeVirtualModel(candidate) && !isRouterModel(candidate) && !getRoutingRegistry()?.getRouter(candidate.provider));
+}
+/**
+* Physical model of one native-virtual request. `before_provider_request`
+* carries only the payload and `ctx.model` stays virtual, so the payload's
+* dispatched model id is matched against credentialed physical models (a
+* router may only route to those). A unique provider match wins. An id shared
+* by several credentialed providers is used only when every candidate gets the
+* same request treatment (`requestPolicyKey`); the identity then stays
+* ambiguous so response evidence is never pinned to a guessed provider.
+* Differing policies return undefined so identity-dependent request mutations
+* fail closed. Without a credentialed match, the sticky branch candidates and
+* then the full catalog are consulted.
+*/
+function resolveNativeVirtualRequestModel(model, payload, ctx, requestPolicyKey) {
+	if (!isNativeVirtualModel(model)) return void 0;
+	const payloadModelId = getProviderPayloadModelId(payload);
+	if (!payloadModelId) return void 0;
+	const config = readEffectiveCompatConfig();
+	const pick = (matches) => {
+		if (matches.length === 0) return null;
+		const configured = matches.map((candidate) => applyConfiguredTransportToModel(candidate, config));
+		if (new Set(configured.map((candidate) => candidate.provider)).size === 1) return {
+			model: configured[0],
+			identityAmbiguous: false
+		};
+		return (requestPolicyKey ? new Set(configured.map(requestPolicyKey)) : void 0)?.size === 1 ? {
+			model: configured[0],
+			identityAmbiguous: true
+		} : void 0;
+	};
+	try {
+		const credentialed = pick(physicalModelsWithId(ctx?.modelRegistry?.getAvailable?.(), payloadModelId));
+		if (credentialed !== null) return credentialed;
+		const { latestSuccessful, latestAny } = findNativeVirtualDispatches(ctx);
+		for (const dispatch of [latestSuccessful, latestAny]) if (dispatch?.id === payloadModelId) return {
+			model: nativeVirtualDispatchToModel(dispatch, ctx),
+			identityAmbiguous: false
+		};
+		return pick(physicalModelsWithId(ctx?.modelRegistry?.getAll?.(), payloadModelId)) ?? void 0;
+	} catch {
+		return;
+	}
+}
+function describeNativeVirtualRouteNote(selected, effective) {
+	if (!selected || !isNativeVirtualModel(selected)) return void 0;
+	if (!effective || isNativeVirtualModel(effective)) return void 0;
+	return `🔀 Native virtual model ${modelKey(selected)} → latest routed physical model ${modelKey(effective)}. The diagnostics below apply to that physical model; Pi may route later requests elsewhere.`;
 }
 function isCacheHintsServiceV1(value) {
 	const record = asRecord(value);
@@ -1152,8 +1317,15 @@ function isEnabledEnv(value) {
 	const normalized = value.trim().toLowerCase();
 	return normalized === "1" || normalized === "true" || normalized === "yes" || normalized === "on";
 }
+function featureEnabled(feature, envName, defaultValue, env = process.env, config = persistedCacheOptimizerConfig) {
+	const configured = config.features?.[feature];
+	if (configured !== void 0) return configured;
+	if (feature === "virtualRewrite") return isEnabledEnv(env[envName]);
+	if (feature === "promptRewrite" || feature === "skillCompression" || feature === "openAICacheKey") return !isEnabledEnv(env[envName]);
+	return isEnabledEnv(env[envName]) || defaultValue;
+}
 function isToolOrderEnabled(env = process.env) {
-	return runtimeOptimizerEnabled && isEnabledEnv(env[TOOL_ORDER_ENV]);
+	return runtimeOptimizerEnabled && featureEnabled("toolOrder", TOOL_ORDER_ENV, false);
 }
 function parseFooterStatsMode(value) {
 	if (typeof value !== "string") return void 0;
@@ -1171,12 +1343,35 @@ const CACHE_OPTIMIZER_COMMANDS = [
 	"fix",
 	"rollback"
 ];
-const CACHE_OPTIMIZER_CONFIG_ARGUMENTS = ["footer-mode"];
+const CACHE_OPTIMIZER_CONFIG_ARGUMENTS = [
+	"footer-mode",
+	"prompt-rewrite",
+	"virtual-rewrite",
+	"skill-compression",
+	"openai-cache-key",
+	"tool-order",
+	"reset"
+];
+const CACHE_OPTIMIZER_FEATURE_COMMANDS = [
+	"prompt-rewrite",
+	"virtual-rewrite",
+	"skill-compression",
+	"openai-cache-key",
+	"tool-order"
+];
+const CACHE_OPTIMIZER_FEATURE_VALUES = ["on", "off"];
 const CACHE_OPTIMIZER_FOOTER_MODES = [
 	"total",
 	"session",
 	"process"
 ];
+const FEATURE_COMMAND_MAP = {
+	"prompt-rewrite": "promptRewrite",
+	"virtual-rewrite": "virtualRewrite",
+	"skill-compression": "skillCompression",
+	"openai-cache-key": "openAICacheKey",
+	"tool-order": "toolOrder"
+};
 const CACHE_OPTIMIZER_STATS_ARGUMENTS = ["all", "contributors"];
 const CACHE_OPTIMIZER_FIX_ARGUMENTS = ["prompt-cache-key"];
 function filterCommandCompletionItems(values, prefix, argumentPath = "") {
@@ -1210,15 +1405,21 @@ function getCacheOptimizerArgumentCompletions(argumentPrefix) {
 		return filterCommandCompletionItems(CACHE_OPTIMIZER_CONFIG_ARGUMENTS, parts[1], "config");
 	}
 	if (parts.length === 3 && parts[1].toLowerCase() === "footer-mode") return filterCommandCompletionItems(CACHE_OPTIMIZER_FOOTER_MODES, parts[2], "config footer-mode");
+	if (parts.length === 3 && CACHE_OPTIMIZER_FEATURE_COMMANDS.includes(parts[1])) return filterCommandCompletionItems(CACHE_OPTIMIZER_FEATURE_VALUES, parts[2], `config ${parts[1]}`);
 	return null;
 }
 function parsePersistedCacheOptimizerConfig(value) {
 	const record = asRecord(value);
-	if (!record || record.version !== 1 && record.version !== 2) return void 0;
-	const allowedTopLevel = new Set(record.version === 1 ? ["version", "footerMode"] : [
+	if (!record || record.version !== 1 && record.version !== 2 && record.version !== 3) return void 0;
+	const allowedTopLevel = new Set(record.version === 1 ? ["version", "footerMode"] : record.version === 2 ? [
 		"version",
 		"footerMode",
 		"promptCacheKey"
+	] : [
+		"version",
+		"footerMode",
+		"promptCacheKey",
+		"features"
 	]);
 	if (Object.keys(record).some((key) => !allowedTopLevel.has(key))) return void 0;
 	const footerMode = parseFooterStatsMode(record.footerMode);
@@ -1227,6 +1428,17 @@ function parsePersistedCacheOptimizerConfig(value) {
 		version: 1,
 		...footerMode ? { footerMode } : {}
 	};
+	const rawFeatures = record.version === 3 ? record.features : void 0;
+	if (rawFeatures !== void 0 && !asRecord(rawFeatures)) return void 0;
+	const featuresRecord = asRecord(rawFeatures);
+	const featureNames = /* @__PURE__ */ new Set([
+		"promptRewrite",
+		"virtualRewrite",
+		"skillCompression",
+		"openAICacheKey",
+		"toolOrder"
+	]);
+	if (featuresRecord && Object.keys(featuresRecord).some((key) => !featureNames.has(key) || typeof featuresRecord[key] !== "boolean")) return void 0;
 	const rawPromptCacheKey = record.promptCacheKey;
 	if (rawPromptCacheKey !== void 0 && !asRecord(rawPromptCacheKey)) return void 0;
 	const promptCacheKey = asRecord(rawPromptCacheKey);
@@ -1236,16 +1448,20 @@ function parsePersistedCacheOptimizerConfig(value) {
 	const stringOmit = omit;
 	const uniqueOmit = stringOmit ? [...new Set(stringOmit.map((value) => value.trim()))].sort() : void 0;
 	return {
-		version: 2,
+		version: record.version === 3 ? 3 : 2,
 		...footerMode ? { footerMode } : {},
-		...uniqueOmit && uniqueOmit.length > 0 ? { promptCacheKey: { omit: uniqueOmit } } : {}
+		...uniqueOmit && uniqueOmit.length > 0 ? { promptCacheKey: { omit: uniqueOmit } } : {},
+		...featuresRecord ? { features: Object.fromEntries(Object.entries(featuresRecord)) } : {}
 	};
 }
 function normalizePersistedCacheOptimizerConfig(value) {
 	if (!value) return { version: 2 };
-	return value.version === 2 ? value : {
-		version: 2,
-		...value.footerMode ? { footerMode: value.footerMode } : {}
+	const features = value.version === 3 ? value.features : void 0;
+	return {
+		version: features ? 3 : 2,
+		...value.footerMode ? { footerMode: value.footerMode } : {},
+		...value.version !== 1 && value.promptCacheKey ? { promptCacheKey: value.promptCacheKey } : {},
+		...features ? { features } : {}
 	};
 }
 function readPersistedCacheOptimizerConfig(configPath = CONFIG_FILE_PATH) {
@@ -1295,7 +1511,7 @@ async function writePersistedFooterModeUnlocked(mode, configPath) {
 		raw = parsePersistedCacheOptimizerConfig(JSON.parse((0, node_fs.readFileSync)(configPath, "utf8")));
 		targetExists = true;
 		if (!raw) throw new Error("invalid footer config schema");
-		version = raw.version === 2 ? 2 : 1;
+		version = raw.version === 3 ? 3 : raw.version === 2 ? 2 : 1;
 	} catch (error) {
 		if (getErrorCode(error) !== "ENOENT") throw new Error("invalid footer config schema");
 	}
@@ -1320,13 +1536,25 @@ async function writePersistedFooterModeUnlocked(mode, configPath) {
 	}
 	await writePersistedCacheOptimizerConfigUnlocked({
 		...current,
-		version: 2,
+		version: 3,
 		footerMode: mode
 	}, configPath);
 }
 async function writePersistedFooterMode(mode, configPath = CONFIG_FILE_PATH) {
 	await withModelsJsonTransactionLock(() => writePersistedFooterModeUnlocked(mode, configPath));
 }
+async function writePersistedFeature(feature, enabled) {
+	const current = readPersistedCacheOptimizerConfig();
+	await writePersistedCacheOptimizerConfig({
+		...current,
+		version: 3,
+		features: {
+			...current.features ?? {},
+			[feature]: enabled
+		}
+	});
+	setPersistedCacheOptimizerConfig(readPersistedCacheOptimizerConfig());
+}
 function configReceiptBackupPath(receipt, receiptPath = CONFIG_RECEIPT_PATH) {
 	return (0, node_path.join)((0, node_path.dirname)(receiptPath), receipt.backupFile);
 }
@@ -1706,7 +1934,7 @@ function isDisabledEnv(value) {
 }
 function shouldInjectOpenAIPromptCacheKey() {
 	if (!runtimeOptimizerEnabled) return false;
-	if (isEnabledEnv(process.env[NO_OPENAI_CACHE_KEY_ENV])) return false;
+	if (!featureEnabled("openAICacheKey", NO_OPENAI_CACHE_KEY_ENV, true)) return false;
 	if (isDisabledEnv(process.env[OPENAI_CACHE_KEY_ENV])) return false;
 	return true;
 }
@@ -1722,16 +1950,60 @@ function getOptimizerRuntimeModeLines() {
 	const state = runtimeOptimizerEnabled ? "enabled" : "disabled";
 	const lines = [];
 	lines.push(`Runtime state: ${state}`);
-	lines.push(`• Prompt rewrite: ${runtimeOptimizerEnabled && !isEnabledEnv(process.env[NO_PROMPT_REWRITE_ENV]) ? "on" : "off"}`);
+	lines.push(`• Prompt rewrite: ${runtimeOptimizerEnabled && featureEnabled("promptRewrite", NO_PROMPT_REWRITE_ENV, true) ? "on" : "off"}`);
+	lines.push(`• Native virtual rewrite: ${featureEnabled("virtualRewrite", VIRTUAL_REWRITE_ENV, false) ? "opt-in" : "off"}`);
+	lines.push(`• Skill compression: ${featureEnabled("skillCompression", NO_SKILL_COMPRESSION_ENV, true) ? "on" : "off"}`);
 	lines.push(`• Deterministic tool ordering: ${isToolOrderEnabled() ? "on (verified built-in payloads, opt-in)" : "off"}`);
 	lines.push(`• OpenAI prompt_cache_key fallback: ${shouldInjectOpenAIPromptCacheKey() ? "on" : "off"}`);
 	lines.push(`• Footer cache stats: on${runtimeOptimizerEnabled ? "" : " (comparison mode)"}`);
 	lines.push(`• Compat warnings: ${runtimeOptimizerEnabled ? "on" : "off"}`);
 	lines.push(`• ${PI_CACHE_RETENTION_ENV}: ${process.env[PI_CACHE_RETENTION_ENV] ?? "(unset)"}`);
+	lines.push(`• Persistent feature overrides: ${Object.keys(persistedCacheOptimizerConfig.features ?? {}).length > 0 ? "configured" : "none"}`);
 	if (!runtimeOptimizerEnabled) lines.push("This is a current-process switch. Run /reload or restart Pi to return to startup behavior.");
-	else if (isEnabledEnv(process.env[NO_PROMPT_REWRITE_ENV]) || !shouldInjectOpenAIPromptCacheKey()) lines.push("Some features are still disabled by environment variables.");
+	else if (!featureEnabled("promptRewrite", NO_PROMPT_REWRITE_ENV, true) || !shouldInjectOpenAIPromptCacheKey()) lines.push("Some features are still disabled by environment variables.");
 	return lines;
 }
+function formatPersistentFeatureConfig() {
+	const features = persistedCacheOptimizerConfig.features ?? {};
+	const lines = ["Persistent feature configuration:"];
+	const entries = [
+		[
+			"Prompt rewrite",
+			"promptRewrite",
+			NO_PROMPT_REWRITE_ENV,
+			true
+		],
+		[
+			"Native virtual rewrite",
+			"virtualRewrite",
+			VIRTUAL_REWRITE_ENV,
+			false
+		],
+		[
+			"Skill compression",
+			"skillCompression",
+			NO_SKILL_COMPRESSION_ENV,
+			true
+		],
+		[
+			"OpenAI cache key",
+			"openAICacheKey",
+			NO_OPENAI_CACHE_KEY_ENV,
+			true
+		],
+		[
+			"Tool ordering",
+			"toolOrder",
+			TOOL_ORDER_ENV,
+			false
+		]
+	];
+	for (const [label, feature, envName, defaultValue] of entries) {
+		const source = features[feature] !== void 0 ? "config" : process.env[envName] !== void 0 ? "env" : "default";
+		lines.push(`• ${label}: ${featureEnabled(feature, envName, defaultValue) ? "on" : "off"} (${source})`);
+	}
+	return lines.join("\n");
+}
 function formatOptimizerRuntimeMode() {
 	return getOptimizerRuntimeModeLines().join("\n");
 }
@@ -1776,14 +2048,14 @@ function isOpenAICompatibleProxyApi(api) {
 function isPiBuiltInLlamaCppModel(model) {
 	if (lower(model?.provider) !== "llama.cpp" || !isOpenAICompatibleProxyApi(model?.api)) return false;
 	const compat = getCompat(model);
-	return compat.supportsStore === false && compat.supportsDeveloperRole === false && compat.supportsReasoningEffort === false && compat.supportsUsageInStreaming === false && compat.supportsStrictMode === false && compat.maxTokensField === "max_tokens" && compat.sendSessionAffinityHeaders === void 0 && compat.sessionAffinityFormat === void 0 && compat.supportsLongCacheRetention === void 0;
+	return compat.supportsStore === false && compat.supportsDeveloperRole === false && compat.supportsReasoningEffort === false && compat.supportsStrictMode === false && compat.maxTokensField === "max_tokens" && compat.sendSessionAffinityHeaders === void 0 && compat.sessionAffinityFormat === void 0 && compat.supportsLongCacheRetention === void 0;
 }
 function shouldInjectOpenAIPromptCacheKeyForModel(model) {
 	return isOpenAICompatibleApi(model?.api);
 }
 function isPromptCacheKeyOmittedForModel(model, config = persistedCacheOptimizerConfig) {
 	if (!model || !isOpenAICompatibleApi(model.api)) return false;
-	return config.promptCacheKey?.omit?.includes(modelKey(model)) === true;
+	return "promptCacheKey" in config && config.promptCacheKey?.omit?.includes(modelKey(model)) === true;
 }
 function setPersistedCacheOptimizerConfig(config) {
 	persistedCacheOptimizerConfig = normalizePersistedCacheOptimizerConfig(config);
@@ -2512,6 +2784,13 @@ function snapshotProviderRequestModel(model) {
 		maxTokens: model.maxTokens ?? 0
 	};
 }
+const MAX_PROVIDER_REQUEST_STATES = 32;
+function pruneProviderRequestStates(states, max = MAX_PROVIDER_REQUEST_STATES) {
+	while (states.length > max) {
+		const completed = states.findIndex((state) => state.responseReceived);
+		states.splice(completed >= 0 ? completed : 0, 1);
+	}
+}
 function keyForModelExt(model) {
 	return `${model.provider}/${model.id}`;
 }
@@ -3998,6 +4277,7 @@ const CACHE_PROVIDER_ADAPTERS = [
 	}
 ];
 function selectAdapterForModel(model) {
+	if (isNativeVirtualModel(model)) return void 0;
 	return CACHE_PROVIDER_ADAPTERS.find((adapter) => adapter.matchesModel(model));
 }
 function selectAdapterForAssistantMessage(message, model) {
@@ -4959,6 +5239,7 @@ function describeRouterChannelDiagnostics(model) {
 }
 function getCompatCheckNotApplicableLines(model) {
 	const api = lower(model.api);
+	if (isNativeVirtualModel(model)) return ["ℹ️ Compat check not applicable for this model.", "   Native virtual model: Pi routes each request to a physical model and none has answered on this session branch yet. Send a prompt, then rerun to diagnose the physical model that answered."];
 	if (isMistralConversationsApi(api)) return ["ℹ️ Compat check not applicable for this model.", "   Native Mistral `mistral-conversations` uses provider-native transport; OpenAI-compatible proxy compat flags do not apply."];
 	if (api === "azure-openai-responses") return ["ℹ️ Compat check not applicable for this model.", "   Native Azure OpenAI Responses uses the Responses transport; OpenAI-compatible proxy compat flags do not apply."];
 	if (api === "openai-codex-responses" || api === "openai-responses" && isOfficialOpenAIBaseUrl(model)) return ["ℹ️ Compat check not applicable for this model.", "   Native Responses transports already use Pi core request handling; OpenAI-compatible proxy compat flags do not apply."];
@@ -7135,6 +7416,8 @@ const __internals_for_tests = {
 	SKILL_COMPRESSION_MIN_COUNT,
 	NO_PROMPT_REWRITE_ENV,
 	isEnabledEnv,
+	featureEnabled,
+	writePersistedFeature,
 	addOpenAIPromptCacheKey,
 	clampPromptCacheKey,
 	hasEffectivePromptCacheKey,
@@ -7152,6 +7435,8 @@ const __internals_for_tests = {
 	normalizeAnthropicCacheControlTtlOrder,
 	isPiBuiltInLlamaCppModel,
 	isResponsesPromptRewriteBypassApi,
+	canRewriteNativeVirtualPrompt,
+	VIRTUAL_REWRITE_ENV,
 	isMistralConversationsApi,
 	isOpenAIFamilyModel,
 	isOpenAIFamilyAssistantMessage,
@@ -7392,6 +7677,16 @@ const __internals_for_tests = {
 	applyConfiguredTransportToModel,
 	resolveRouteModel,
 	isVirtualRoutingModel,
+	PI_VIRTUAL_MODEL_API,
+	isNativeVirtualModel,
+	findNativeVirtualDispatches,
+	resolveNativeVirtualRouteModel,
+	resolveNativeVirtualRequestModel,
+	getProviderPayloadModelId,
+	getCompatCheckNotApplicableLines,
+	describeNativeVirtualRouteNote,
+	pruneProviderRequestStates,
+	MAX_PROVIDER_REQUEST_STATES,
 	installCacheHintsService,
 	getCacheHintsService,
 	markOptimizerOwnedCacheHintsService,
@@ -7895,6 +8190,23 @@ function pi_cache_optimizer_default(pi) {
 	function hasExplicitLongRetentionOptIn(model) {
 		return hasExplicitLongRetentionOptInFromConfig(readEffectiveCompatConfig(), model.provider, model.id);
 	}
+	/**
+	* Request treatment that before_provider_request derives from a model's
+	* identity. A native virtual request whose dispatched id matches several
+	* credentialed providers is resolved only when this key agrees for all of
+	* them, so no provider-specific decision is applied on a guess.
+	*/
+	function requestPolicyKey(model) {
+		const key = modelKey(model);
+		return JSON.stringify([
+			model.api ?? "",
+			isOfficialOpenAIBaseUrl(model),
+			hasExplicitLongRetentionOptIn(model),
+			promptCacheRetention400Models.has(key),
+			anthropicTtlOrderErrorModels.has(key),
+			isPromptCacheKeyOmittedForModel(model)
+		]);
+	}
 	pi.on("session_start", async (event, ctx) => {
 		if (runtimeOptimizerEnabled) requestLongCacheRetention();
 		await restoreCacheStats(event.reason, ctx);
@@ -7904,7 +8216,8 @@ function pi_cache_optimizer_default(pi) {
 		} catch {}
 		await publishStatus(ctx);
 	});
-	pi.on("tool_execution_end", async (_event, ctx) => {
+	pi.on("tool_execution_end", async (event, ctx) => {
+		if (isNonEmptyString(event?.parentToolCallId)) return;
 		await refreshShardAggregate();
 		await publishStatus(ctx);
 	});
@@ -7938,8 +8251,9 @@ function pi_cache_optimizer_default(pi) {
 		const routeSnapshot = resolveActiveRouteSnapshot(_ctx.model, _ctx);
 		const model = (routeSnapshot ? findModelInRegistry(_ctx.modelRegistry, routeSnapshot.provider, routeSnapshot.modelId) ?? routeSnapshotToPiModel(routeSnapshot, _ctx.model) : void 0) ?? _ctx.model;
 		if (model && isResponsesPromptRewriteBypassApi(model.api)) return {};
+		if (isNativeVirtualModel(_ctx.model) && !canRewriteNativeVirtualPrompt(_ctx.model, _ctx)) return {};
 		if (!runtimeOptimizerEnabled) return {};
-		if (isEnabledEnv(process.env[NO_PROMPT_REWRITE_ENV])) return {};
+		if (!featureEnabled("promptRewrite", NO_PROMPT_REWRITE_ENV, true)) return {};
 		const strippedPrompt = stripSessionOverviewChurn(event.systemPrompt);
 		const compressedPrompt = compressSkillsInSystemPrompt(strippedPrompt, event.systemPromptOptions);
 		const optimized = optimizeSystemPrompt(compressedPrompt, event.systemPromptOptions);
@@ -7978,17 +8292,23 @@ function pi_cache_optimizer_default(pi) {
 		return {};
 	});
 	pi.on("before_provider_headers", (event, ctx) => {
+		if (isNativeVirtualModel(ctx.model)) return;
 		const requestModel = resolveRouteModel(ctx.model, ctx) ?? ctx.model;
 		addEffectiveSessionAffinityHeaders(event.headers, requestModel, ctx.sessionManager.getSessionId());
 	});
 	pi.on("before_provider_request", (event, ctx) => {
-		const requestModel = resolveRouteModel(ctx.model, ctx) ?? ctx.model;
+		const nativeVirtualRequest = resolveNativeVirtualRequestModel(ctx.model, event.payload, ctx, requestPolicyKey);
+		const requestModel = isNativeVirtualModel(ctx.model) ? nativeVirtualRequest?.model ?? ctx.model : resolveRouteModel(ctx.model, ctx) ?? ctx.model;
 		const snapshot = snapshotProviderRequestModel(requestModel);
-		if (snapshot) providerRequestStates.push({
-			model: snapshot,
-			responseReceived: false,
-			correlationAmbiguous: false
-		});
+		if (snapshot) {
+			providerRequestStates.push({
+				model: snapshot,
+				responseReceived: false,
+				correlationAmbiguous: false,
+				identityAmbiguous: isNativeVirtualModel(ctx.model) && (!nativeVirtualRequest || nativeVirtualRequest.identityAmbiguous)
+			});
+			pruneProviderRequestStates(providerRequestStates);
+		}
 		let requestPayload = event.payload;
 		let toolOrderChanged = false;
 		if (isToolOrderEnabled() && isToolOrderingEligibleModel(requestModel)) {
@@ -8021,7 +8341,7 @@ function pi_cache_optimizer_default(pi) {
 			else for (const state of pendingStates) state.correlationAmbiguous = true;
 		}
 		if (responseState) responseState.responseReceived = true;
-		const model = responseState?.model ?? (pendingStates.length === 0 ? resolveRouteModel(ctx.model, ctx) ?? ctx.model : void 0);
+		const model = responseState ? responseState.identityAmbiguous ? void 0 : responseState.model : pendingStates.length === 0 && !isNativeVirtualModel(ctx.model) ? resolveRouteModel(ctx.model, ctx) ?? ctx.model : void 0;
 		if (!runtimeOptimizerEnabled || !model) return;
 		if (isReasoningProtocolRejectionSignalApplicable(model) && event.status === 400 && hasReasoningProtocolRejectionSignal(event.headers)) await notifyReasoningProtocolObservation(model, ctx, reasoningProtocolRejectedModels, warnedReasoningProtocolRejectedModels);
 		if (event.status === 400 && isPromptCacheKeyUnsupportedApplicable(model) && hasPromptCacheKeyUnsupportedSignal(event.headers)) {
@@ -8073,9 +8393,10 @@ function pi_cache_optimizer_default(pi) {
 				model: void 0,
 				ambiguous: false
 			};
+			const ambiguous = (state.correlationAmbiguous || state.identityAmbiguous === true) && explicitIndex < 0;
 			return {
-				model: state.correlationAmbiguous && explicitIndex < 0 ? void 0 : state.model,
-				ambiguous: state.correlationAmbiguous && explicitIndex < 0
+				model: ambiguous ? void 0 : state.model,
+				ambiguous
 			};
 		})() : {
 			model: void 0,
@@ -8127,12 +8448,15 @@ function pi_cache_optimizer_default(pi) {
 				}
 			}
 		}
-		const adapter = selectAdapterForAssistantMessage(event.message, ctx.model);
+		const nativeVirtualDispatch = isNativeVirtualModel(ctx.model) ? nativeVirtualDispatchFromMessage(event.message) : void 0;
+		const nativeVirtualMessageModel = nativeVirtualDispatch ? nativeVirtualDispatchToModel(nativeVirtualDispatch, ctx) : void 0;
+		const adapter = selectAdapterForAssistantMessage(event.message, nativeVirtualMessageModel ?? ctx.model);
 		if (!adapter) return;
 		if (msgRecord?.stopReason === "error" || msgRecord?.stopReason === "aborted") return;
 		const usage = adapter.normalizeUsage(event.message);
 		let statsModel = modelFromAssistantMessage(event.message, ctx.model) ?? ctx.model;
 		statsModel = consolidateDirectProviderStatsModel(statsModel, ctx.model, ctx);
+		if (nativeVirtualMessageModel) statsModel = nativeVirtualMessageModel;
 		let routedModelChanged = false;
 		if (isVirtualRoutingModel(ctx.model, ctx) && statsModel && !isVirtualRoutingModel(statsModel, ctx)) {
 			const nextRoutedModel = {
@@ -8217,7 +8541,11 @@ function pi_cache_optimizer_default(pi) {
 			const adapter = selectAdapterForModel(model);
 			const sk = model ? sessionModelKey(model) : void 0;
 			const lowHitLines = buildLowHitDiagnosis(model, adapter, model ? cacheStatsTotalsByModel[modelKey(model)] : void 0, sk ? getRecentSamples(sk) : []);
-			const fullDiagnosis = lowHitLines.length > 0 ? diagnosis + "\n" + lowHitLines.join("\n") : diagnosis;
+			const fullDiagnosis = [
+				describeNativeVirtualRouteNote(selectedModel, model),
+				diagnosis,
+				...lowHitLines
+			].filter((line) => line !== void 0).join("\n");
 			cmdCtx.ui.notify(fullDiagnosis, "info");
 		} else if (subcommand === "stats") {
 			const aggregate = await refreshShardAggregate();
@@ -8232,13 +8560,48 @@ function pi_cache_optimizer_default(pi) {
 		} else if (subcommand === "config") {
 			const configKey = commandParts[1];
 			const requestedMode = commandParts[2];
+			const feature = configKey ? FEATURE_COMMAND_MAP[configKey] : void 0;
+			if (feature && (requestedMode === "on" || requestedMode === "off")) {
+				try {
+					await writePersistedFeature(feature, requestedMode === "on");
+					lastStatusText = void 0;
+					await publishStatus(cmdCtx, model);
+					cmdCtx.ui.notify(`✅ ${configKey} set to ${requestedMode}. Persistent config overrides its environment variable.`, "info");
+				} catch (error) {
+					cmdCtx.ui.notify(`❌ Could not update ${configKey}: ${error instanceof Error ? error.message : String(error)}`, "error");
+				}
+				return;
+			}
+			if (configKey === "reset") {
+				try {
+					await writePersistedCacheOptimizerConfig({
+						version: 2,
+						footerMode: persistedFooterStatsMode,
+						promptCacheKey: persistedCacheOptimizerConfig.promptCacheKey
+					});
+					setPersistedCacheOptimizerConfig(readPersistedCacheOptimizerConfig());
+					cmdCtx.ui.notify("✅ Feature configuration reset. Environment variables now apply again.", "info");
+				} catch (error) {
+					cmdCtx.ui.notify(`❌ Could not reset feature configuration: ${error instanceof Error ? error.message : String(error)}`, "error");
+				}
+				return;
+			}
+			if (!configKey) {
+				cmdCtx.ui.notify(formatPersistentFeatureConfig() + `\n• Footer mode: ${resolveFooterStatsMode(persistedFooterStatsMode).mode}
+
+Usage: /cache-optimizer config <feature> on|off | footer-mode total|session|process | reset`, "info");
+				return;
+			}
 			if (configKey !== "footer-mode" || !requestedMode || ![
 				"session",
 				"total",
 				"process"
 			].includes(requestedMode)) {
 				const resolved = resolveFooterStatsMode(persistedFooterStatsMode);
-				cmdCtx.ui.notify(`Usage: /cache-optimizer config footer-mode total|session|process\nCurrent footer mode: ${resolved.mode} (${resolved.source})`, "info");
+				cmdCtx.ui.notify(`Usage: /cache-optimizer config footer-mode total|session|process
+       /cache-optimizer config prompt-rewrite|virtual-rewrite|skill-compression|openai-cache-key|tool-order on|off
+       /cache-optimizer config reset
+Current footer mode: ${resolved.mode} (${resolved.source})`, "info");
 				return;
 			}
 			const nextMode = requestedMode;
@@ -8258,8 +8621,10 @@ function pi_cache_optimizer_default(pi) {
 				return;
 			}
 			const compatResult = buildCompatDiagnosis(model);
-			if (compatResult) cmdCtx.ui.notify(compatResult, "warning");
-			else cmdCtx.ui.notify(isAdaptiveThinkingCompatApplicable(model) || isDeepSeekCompatCheckApplicable(model) || isCompatCheckApplicable(model) ? "✅ Compat fully configured." : getCompatCheckNotApplicableLines(model).join("\n"), "info");
+			const compatRouteNote = describeNativeVirtualRouteNote(selectedModel, model);
+			const withRouteNote = (text) => compatRouteNote ? `${compatRouteNote}\n${text}` : text;
+			if (compatResult) cmdCtx.ui.notify(withRouteNote(compatResult), "warning");
+			else cmdCtx.ui.notify(withRouteNote(isAdaptiveThinkingCompatApplicable(model) || isDeepSeekCompatCheckApplicable(model) || isCompatCheckApplicable(model) ? "✅ Compat fully configured." : getCompatCheckNotApplicableLines(model).join("\n")), "info");
 		} else if (subcommand === "rollback") {
 			if (!model) {
 				cmdCtx.ui.notify("No active model selected. Select a model first with /model or pi --model.", "warning");
@@ -8677,7 +9042,9 @@ Run /cache-optimizer fix prompt-cache-key in Pi's interactive UI for ${modelKey(
 			diagnosis.push("  stats all — Show detailed totals for every model across all local sessions");
 			diagnosis.push("  stats contributors — Show per-session contributors for the active model");
 			diagnosis.push("  compat  — Show compat suggestion with edit location");
+			diagnosis.push("  config prompt-rewrite|virtual-rewrite|skill-compression|openai-cache-key|tool-order on|off — Persist feature settings");
 			diagnosis.push("  config footer-mode total|session|process — Persist the footer stats mode");
+			diagnosis.push("  config reset — Remove persistent feature overrides");
 			diagnosis.push("  fix     — Auto-fix compat issues (writes models.json or extension config, requires UI)");
 			diagnosis.push("  fix prompt-cache-key — Explicitly omit prompt_cache_key for the active model");
 			diagnosis.push("  rollback — Undo the latest confirmed fix (requires UI confirmation)");
@@ -8713,4 +9080,4 @@ Run /cache-optimizer fix prompt-cache-key in Pi's interactive UI for ${modelKey(
 exports.__internals_for_tests = __internals_for_tests;
 exports.default = pi_cache_optimizer_default;
 
-//# sourceMappingURL=pi-cache-optimizer-VJG1vKKU.cjs.map
+//# sourceMappingURL=pi-cache-optimizer-DSED41mZ.cjs.map