@vtxmacro/cli 2026.8.38 → 2026.8.39

package/README.md

@@ -58,10 +58,10 @@ exposed. Before starting a Trader, confirm that the VTX AI page shows the
 intended authenticated ChatGPT email and plan. The host reads the live Codex
 account window: a reached limit pauses new dispatch until its reported reset,
 while a transient throttle uses a short bounded cooldown.
-The selected AI model row shows the open VTX profile's rolling 1-hour, 24-hour,
-and 7-day request and reported-token activity for that subscription path.
-Account-wide quota percentages remain in safe host diagnostics instead of the
-profile-scoped selector.
+The selected AI model row estimates the open VTX profile's attributable
+membership burn rate for that subscription, expressed as equivalent percentage
+points per hour, day, and seven days. Account-wide quota percentages remain in
+safe host diagnostics instead of the profile-scoped selector.
 
 ```bash
 vtx inference-host login
@@ -86,12 +86,12 @@ distinct. Use `service uninstall --instance <name>` to remove one worker;
 unqualified `service uninstall` removes the whole supervisor. The legacy
 unqualified host commands continue to target `default`.
 
-For one VTX account and adapter, the selected host is tried first and other
-compatible memberships follow in setup order. Quota/credits exhaustion,
-unusable authentication, and exhausted recoverable provider failures can
-advance the same logical call. VTX never changes provider, model, effort, or
-response mode, and never cascades a bad request, policy rejection, or uncertain
-dispatch outcome.
+For each VTX profile and lane, **Subscription #1** is tried first and later
+compatible subscriptions follow in the saved Account cascade order.
+Quota/credits exhaustion, unusable authentication, and exhausted recoverable
+provider failures can advance the same logical call. VTX never changes
+provider, model, effort, or response mode, and never cascades a bad request,
+policy rejection, or uncertain dispatch outcome.
 
 Automated hosts do not impose a subscription-specific profile-count or
 concurrency limit by default. Set an explicit positive integer with

package/bin/vtx.js

@@ -38,7 +38,7 @@ var init_agent_cli_release = __esm({
   "agent-cli-release.json"() {
     agent_cli_release_default = {
       package_name: "@vtxmacro/cli",
-      package_version: "2026.8.38",
+      package_version: "2026.8.39",
       codex_package_name: "@openai/codex",
       codex_version: "0.147.0",
       platforms: {
@@ -14375,7 +14375,7 @@ var init_zod = __esm({
 });
 
 // lib/external-inference-contract.ts
-var EXTERNAL_INFERENCE_CONTRACT_VERSION, EXTERNAL_INFERENCE_CONTROLLER_SCHEMA_VERSION, EXTERNAL_INFERENCE_ADVERTISEMENT_SCHEMA_VERSION, EXTERNAL_INFERENCE_ENVELOPE_SCHEMA_VERSION, EXTERNAL_INFERENCE_JOB_INPUT_SCHEMA_VERSION, LEGACY_EXTERNAL_INFERENCE_JOB_INPUT_SCHEMA_VERSION, EXTERNAL_INFERENCE_USAGE_SCHEMA_VERSION, EXTERNAL_INFERENCE_INPUT_HASH_DOMAIN, MAX_EXTERNAL_INFERENCE_STATUS_HOSTS, MAX_GENERATION, MAX_SAFE_INTEGER, MAX_PROMPT_BYTES, MAX_IMMUTABLE_JOB_INPUT_BYTES, MAX_CANDIDATE_PLAINTEXT_BYTES, MAX_CIPHERTEXT_CHARS, SHA256_INITIAL_STATE, SHA256_ROUND_CONSTANTS, rotateRight, sha256BytesHex, sha256Utf8Hex, computeImmutableJobInputSha256, identifierSchema, operationIdentifierSchema, sha256HexSchema, generationSchema, zeroBasedGenerationSchema, nonNegativeSafeIntegerSchema, positiveSafeIntegerSchema, sqlIntegerTokenCountSchema, timestampSchema, safeCodeSchema, externalInferenceAdapterIdSchema, reasoningEffortSchema, modelNameSchema, displayNameSchema, authenticatedAccountEmailSchema, protocolVersionSchema, base64UrlAlphabet, isCanonicalBase64Url, base64Url32BytesSchema, base64UrlNonceSchema, base64UrlCiphertextSchema, claimHandleSchema, boundedJsonTextSchema, boundedPromptTextSchema, laneSchema, botModeSchema, executionModeSchema, responseModeSchema, dispatchOutcomeSchema, responseOutcomeSchema, nativeModelControllerSchema, externalAgentModelControllerSchema, externalAgentUnboundControllerSchema, liveModelControllerSchema, snapshotModelControllerSchema, modelIdentityMappingSchema, advertisedModelSchema, hostAdvertisementSchema, envelopeAadSchema, validateEnvelopeAad, commonEnvelopeShape, sealedKeyWrapEnvelopeSchema, sealedInputEnvelopeSchema, sealedCandidateEnvelopeSchema, sealedEnvelopeSchema, screenerCandidateRuntimeBindingSchema, screenerClientLeaseBindingSchema, screenerRuntimeBindingSchema, serverRuntimeBindingSchema, clientRuntimeBindingSchema, runtimeBindingSchema, immutableJobInputSchema, canonicalizeJsonValue, serializeImmutableJobInput, reportedTokenUsageSchema, unavailableTokenUsageSchema, tokenUsageSchema, agentConnectRequestSchema, agentNextRequestSchema, agentHeartbeatRequestSchema, emptyAgentNextResultSchema, claimedAgentNextResultSchema, agentNextResultSchema, agentCompleteRequestSchema, agentFailRequestSchema, requestedModelSelectionSchema, completedModelProvenanceSchema, attemptModelProvenanceSchema, providerWeeklyQuotaReadSchema, hostHeartbeatRequestSchema, jobClaimRequestSchema, emptyJobClaimSchema, claimedJobSchema, jobClaimResultSchema, attemptStartRequestSchema, jobHeartbeatRequestSchema, continueHeartbeatDirectiveSchema, cancelHeartbeatDirectiveSchema, jobHeartbeatResultSchema, adapterOutcomeEvidenceSchema, completionBaseShape, validateCompletion, providerResponseCompletionSchema, decisionCandidateCompletionSchema, jobCompleteRequestSchema, jobFailRequestSchema, jobCancelRequestSchema, jobCancellationReceiptSchema, jobDeliveryRequestSchema, deliveryBaseShape, validateDeliveryCandidate, providerResponseDeliverySchema, decisionCandidateDeliverySchema, jobDeliveryResultSchema, jobConsumptionRequestSchema, jobConsumptionReceiptSchema, jobStateSchema, jobStatusReadSchema, hostStatusReadSchema, controllerBindingStatusReadSchema, queueStatusReadSchema, outcomeStatusReadSchema, externalInferenceVolumeWindowReadSchema, externalInferenceFailureCountReadSchema, externalInferenceHostDiagnosticsReadSchema, externalInferenceProfileActivityReadSchema, externalInferenceStatusReadSchema, nativeResearchEvidenceSourceSchema, externalAgentResearchEvidenceSourceSchema, researchEvidenceSourceSchema, reportedResearchTokenUsageSchema, unavailableResearchTokenUsageSchema, researchTokenUsageSchema, externalInferenceProductLineageBaseShape, traderExternalInferenceProductLineageSchema, decisionChatExternalInferenceProductLineageSchema, screenerExternalInferenceProductLineageSchema, externalInferenceProductLineageSchema, normalProductLinkageSchema, evidenceMaterializationMarkerSchema, deliveredExternalAgentSourceIdentitySchema, deliveredRequestEvidenceSchema, researchEvidenceBaseShape, researchEvidenceBaseSchema, validateExactUtf8, validateResearchEvidence, providerResponseResearchEvidenceSchema, decisionCandidateResearchEvidenceSchema, researchEvidenceSchema;
+var EXTERNAL_INFERENCE_CONTRACT_VERSION, EXTERNAL_INFERENCE_CONTROLLER_SCHEMA_VERSION, EXTERNAL_INFERENCE_ADVERTISEMENT_SCHEMA_VERSION, EXTERNAL_INFERENCE_ENVELOPE_SCHEMA_VERSION, EXTERNAL_INFERENCE_JOB_INPUT_SCHEMA_VERSION, LEGACY_EXTERNAL_INFERENCE_JOB_INPUT_SCHEMA_VERSION, EXTERNAL_INFERENCE_USAGE_SCHEMA_VERSION, EXTERNAL_INFERENCE_INPUT_HASH_DOMAIN, MAX_EXTERNAL_INFERENCE_STATUS_HOSTS, MAX_GENERATION, MAX_SAFE_INTEGER, MAX_PROMPT_BYTES, MAX_IMMUTABLE_JOB_INPUT_BYTES, MAX_CANDIDATE_PLAINTEXT_BYTES, MAX_CIPHERTEXT_CHARS, SHA256_INITIAL_STATE, SHA256_ROUND_CONSTANTS, rotateRight, sha256BytesHex, sha256Utf8Hex, computeImmutableJobInputSha256, identifierSchema, operationIdentifierSchema, sha256HexSchema, generationSchema, zeroBasedGenerationSchema, nonNegativeSafeIntegerSchema, positiveSafeIntegerSchema, sqlIntegerTokenCountSchema, timestampSchema, safeCodeSchema, externalInferenceAdapterIdSchema, reasoningEffortSchema, modelNameSchema, displayNameSchema, authenticatedAccountEmailSchema, protocolVersionSchema, base64UrlAlphabet, isCanonicalBase64Url, base64Url32BytesSchema, base64UrlNonceSchema, base64UrlCiphertextSchema, claimHandleSchema, boundedJsonTextSchema, boundedPromptTextSchema, laneSchema, botModeSchema, executionModeSchema, responseModeSchema, dispatchOutcomeSchema, responseOutcomeSchema, nativeModelControllerSchema, externalAgentModelControllerSchema, externalAgentUnboundControllerSchema, orderedFallbackHostIdsSchema, externalInferenceFallbackHostIdsSchema, liveModelControllerSchema, snapshotModelControllerSchema, modelIdentityMappingSchema, advertisedModelSchema, hostAdvertisementSchema, envelopeAadSchema, validateEnvelopeAad, commonEnvelopeShape, sealedKeyWrapEnvelopeSchema, sealedInputEnvelopeSchema, sealedCandidateEnvelopeSchema, sealedEnvelopeSchema, screenerCandidateRuntimeBindingSchema, screenerClientLeaseBindingSchema, screenerRuntimeBindingSchema, serverRuntimeBindingSchema, clientRuntimeBindingSchema, runtimeBindingSchema, immutableJobInputSchema, canonicalizeJsonValue, serializeImmutableJobInput, reportedTokenUsageSchema, unavailableTokenUsageSchema, tokenUsageSchema, agentConnectRequestSchema, agentNextRequestSchema, agentHeartbeatRequestSchema, emptyAgentNextResultSchema, claimedAgentNextResultSchema, agentNextResultSchema, agentCompleteRequestSchema, agentFailRequestSchema, requestedModelSelectionSchema, completedModelProvenanceSchema, attemptModelProvenanceSchema, providerWeeklyQuotaReadSchema, hostHeartbeatRequestSchema, jobClaimRequestSchema, emptyJobClaimSchema, claimedJobSchema, jobClaimResultSchema, attemptStartRequestSchema, jobHeartbeatRequestSchema, continueHeartbeatDirectiveSchema, cancelHeartbeatDirectiveSchema, jobHeartbeatResultSchema, adapterOutcomeEvidenceSchema, completionBaseShape, validateCompletion, providerResponseCompletionSchema, decisionCandidateCompletionSchema, jobCompleteRequestSchema, jobFailRequestSchema, jobCancelRequestSchema, jobCancellationReceiptSchema, jobDeliveryRequestSchema, deliveryBaseShape, validateDeliveryCandidate, providerResponseDeliverySchema, decisionCandidateDeliverySchema, jobDeliveryResultSchema, jobConsumptionRequestSchema, jobConsumptionReceiptSchema, jobStateSchema, jobStatusReadSchema, hostStatusReadSchema, controllerBindingStatusReadSchema, queueStatusReadSchema, outcomeStatusReadSchema, externalInferenceVolumeWindowReadSchema, externalInferenceFailureCountReadSchema, externalInferenceHostDiagnosticsReadSchema, externalInferenceProfileActivityReadSchema, externalInferenceStatusReadSchema, nativeResearchEvidenceSourceSchema, externalAgentResearchEvidenceSourceSchema, researchEvidenceSourceSchema, reportedResearchTokenUsageSchema, unavailableResearchTokenUsageSchema, researchTokenUsageSchema, externalInferenceProductLineageBaseShape, traderExternalInferenceProductLineageSchema, decisionChatExternalInferenceProductLineageSchema, screenerExternalInferenceProductLineageSchema, externalInferenceProductLineageSchema, normalProductLinkageSchema, evidenceMaterializationMarkerSchema, deliveredExternalAgentSourceIdentitySchema, deliveredRequestEvidenceSchema, researchEvidenceBaseShape, researchEvidenceBaseSchema, validateExactUtf8, validateResearchEvidence, providerResponseResearchEvidenceSchema, decisionCandidateResearchEvidenceSchema, researchEvidenceSchema;
 var init_external_inference_contract = __esm({
   "lib/external-inference-contract.ts"() {
     "use strict";
@@ -14646,6 +14646,15 @@ var init_external_inference_contract = __esm({
       response_mode_hint: responseModeSchema,
       binding_state: external_exports.literal("rebind_required")
     });
+    orderedFallbackHostIdsSchema = external_exports.array(identifierSchema).max(127).refine(
+      (values) => new Set(values).size === values.length,
+      { message: "fallback host IDs must be unique" }
+    );
+    externalInferenceFallbackHostIdsSchema = external_exports.strictObject({
+      main: orderedFallbackHostIdsSchema,
+      review: orderedFallbackHostIdsSchema,
+      screener: orderedFallbackHostIdsSchema
+    });
     liveModelControllerSchema = external_exports.discriminatedUnion("kind", [
       nativeModelControllerSchema,
       externalAgentModelControllerSchema
@@ -19496,7 +19505,7 @@ async function logoutCodexSubscription(options) {
     await session.close();
   }
 }
-var CODEX_INFERENCE_PERMISSION_PROFILE, CODEX_REASONING_CONTENT_MAX_UTF8_BYTES, CODEX_REASONING_SUMMARY_MAX_UTF8_BYTES, CODEX_ACCOUNT_PLAN_TYPES, CodexAppServerError, objectOrNull, finiteToken, tokenUsageFromBreakdown, usageFromNotification, validPlanType, nonnegativeSafeIntegerOrNull, nonnegativeFiniteNumberOrNull, parseRateLimitWindow, RATE_LIMIT_REACHED_TYPES, DEFAULT_CODEX_QUOTA_COOLDOWN_MS, MAX_CODEX_QUOTA_COOLDOWN_MS, CODEX_TRANSIENT_RATE_LIMIT_COOLDOWN_MS, parseRateLimitSnapshot, codexRateLimitRetryAtMs, codexAccountRateLimitReached, forbiddenMethod, forbiddenTerminalItem, classifyCodexTurnFailure, scrubbedCodexEnvironment, killWindowsProcessTree, appServerArgs, GUARDIAN_SCRIPT, WINDOWS_RECEIPT_REPLACE_ERROR_CODES, replaceCodexGuardianReceiptFile, writeCodexGuardianSpawnIntent, parseGuardianReceipt, readCodexGuardianReceipt, waitForCodexGuardianState, defaultSpawn, CodexAppServerSession;
+var CODEX_INFERENCE_PERMISSION_PROFILE, CODEX_REASONING_CONTENT_MAX_UTF8_BYTES, CODEX_REASONING_SUMMARY_MAX_UTF8_BYTES, CODEX_ACCOUNT_PLAN_TYPES, CodexAppServerError, objectOrNull, finiteToken, tokenUsageFromBreakdown, usageFromNotification, validPlanType, nonnegativeSafeIntegerOrNull, nonnegativeFiniteNumberOrNull, parseRateLimitWindow, RATE_LIMIT_REACHED_TYPES, DEFAULT_CODEX_QUOTA_COOLDOWN_MS, MAX_CODEX_QUOTA_COOLDOWN_MS, CODEX_TRANSIENT_RATE_LIMIT_COOLDOWN_MS, parseRateLimitSnapshot, codexRateLimitRetryAtMs, codexAccountRateLimitReached, forbiddenMethod, forbiddenTerminalItem, classifyCodexTurnFailure, scrubbedCodexEnvironment, killWindowsProcessTree, appServerArgs, GUARDIAN_SCRIPT, WINDOWS_RECEIPT_REPLACE_ERROR_CODES, MAX_RETIRED_RESPONSE_IDS, LATE_RESPONSE_SAFE_METHODS, replaceCodexGuardianReceiptFile, writeCodexGuardianSpawnIntent, parseGuardianReceipt, readCodexGuardianReceipt, waitForCodexGuardianState, defaultSpawn, CodexAppServerSession;
 var init_codex_app_server = __esm({
   "lib/inference-host/codex-app-server.ts"() {
     "use strict";
@@ -19959,6 +19968,8 @@ child.once('close', async () => {
       "ENOTEMPTY",
       "EPERM"
     ]);
+    MAX_RETIRED_RESPONSE_IDS = 256;
+    LATE_RESPONSE_SAFE_METHODS = /* @__PURE__ */ new Set(["account/rateLimits/read"]);
     replaceCodexGuardianReceiptFile = async (temporaryPath, receiptPath, options = {}) => {
       const platform = options.platform ?? process.platform;
       const renameFile = options.renameFile ?? rename2;
@@ -20081,6 +20092,7 @@ child.once('close', async () => {
     CodexAppServerSession = class _CodexAppServerSession {
       constructor(options) {
         this.pending = /* @__PURE__ */ new Map();
+        this.retiredResponseIds = /* @__PURE__ */ new Map();
         this.notificationListeners = /* @__PURE__ */ new Set();
         this.fatalListeners = /* @__PURE__ */ new Set();
         this.accountLoginCompletions = /* @__PURE__ */ new Map();
@@ -20219,6 +20231,9 @@ child.once('close', async () => {
         if (typeof id2 === "number" && Number.isSafeInteger(id2) && !("method" in message)) {
           const pending = this.pending.get(id2);
           if (!pending) {
+            const retiredMethod = this.retiredResponseIds.get(id2);
+            this.retiredResponseIds.delete(id2);
+            if (retiredMethod && LATE_RESPONSE_SAFE_METHODS.has(retiredMethod)) return;
             this.failTransport("unexpected_response_id");
             return;
           }
@@ -20283,6 +20298,14 @@ child.once('close', async () => {
         this.pending.clear();
         for (const listener of this.fatalListeners) listener(this.fatalError);
       }
+      retireResponseId(id2, method) {
+        this.retiredResponseIds.set(id2, method);
+        while (this.retiredResponseIds.size > MAX_RETIRED_RESPONSE_IDS) {
+          const oldest = this.retiredResponseIds.keys().next().value;
+          if (typeof oldest !== "number") break;
+          this.retiredResponseIds.delete(oldest);
+        }
+      }
       write(message) {
         if (this.closed || this.fatalError || this.process.exitCode !== null) {
           throw this.fatalError ?? new CodexAppServerError({
@@ -20321,6 +20344,7 @@ child.once('close', async () => {
           const pending = { resolve: resolve6, reject, timer: null, abortCleanup: null };
           pending.timer = setTimeout(() => {
             this.pending.delete(id2);
+            this.retireResponseId(id2, method);
             pending.abortCleanup?.();
             reject(new CodexAppServerError({
               message: "Codex app-server request timed out.",
@@ -20332,6 +20356,7 @@ child.once('close', async () => {
           if (options.signal) {
             const onAbort = () => {
               this.pending.delete(id2);
+              this.retireResponseId(id2, method);
               if (pending.timer) clearTimeout(pending.timer);
               pending.abortCleanup?.();
               reject(new CodexAppServerError({
@@ -22224,9 +22249,21 @@ var init_codex_adapter = __esm({
               dispatchOutcome: checkpoint.dispatchOutcome
             });
           }
-          const receipt = await readCodexGuardianReceipt(checkpoint.processReceiptPath);
-          if (checkpoint.processState === "terminated" || receipt === null || receipt.state === "terminated") {
-            await rm3(checkpoint.processReceiptPath, { force: true });
+          let sharedReceiptStillRequired = true;
+          if (recoveryHooks.loadAll) {
+            try {
+              const checkpoints = await recoveryHooks.loadAll();
+              const receiptPath = resolve3(checkpoint.processReceiptPath);
+              sharedReceiptStillRequired = Object.values(checkpoints).some((candidate) => candidate.attemptId !== attemptId && !candidate.cleanupConfirmed && candidate.processReceiptPath !== null && resolve3(candidate.processReceiptPath) === receiptPath);
+            } catch {
+              sharedReceiptStillRequired = true;
+            }
+          }
+          if (!sharedReceiptStillRequired) {
+            const receipt = await readCodexGuardianReceipt(checkpoint.processReceiptPath);
+            if (checkpoint.processState === "terminated" || receipt === null || receipt.state === "terminated") {
+              await rm3(checkpoint.processReceiptPath, { force: true });
+            }
           }
         }
         await recoveryHooks.clear(attemptId);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@vtxmacro/cli",
-  "version": "2026.8.38",
+  "version": "2026.8.39",
   "description": "VTX Macro CLI, MCP server, and durable subscription inference host.",
   "type": "module",
   "license": "UNLICENSED",