@muggleai/works 5.31.0-staging.271 → 5.31.0-staging.273

package/.claude-plugin/marketplace.json

@@ -9,7 +9,7 @@
       "name": "muggleai",
       "source": "./plugin",
       "description": "Real-browser E2E acceptance for your coding agent: test the flows you changed and open the PR with screenshots. Built for flows other tools can't reach: managed test identities with live inboxes (magic links, email OTP and 2FA, verification mail, password resets), an encrypted secret store for login and test credentials, and a test graph that runs prerequisite flows first. Plain-English tests become replayable scripts on localhost or preview URLs.",
-      "version": "5.31.0-staging.271"
+      "version": "5.31.0-staging.273"
     }
   ]
 }

package/README.md

@@ -273,11 +273,11 @@ so the result carries a viewUrl straight to the dashboard
 
 ## MCP tool reference
 
-125 tools across authentication, projects, use cases, test cases, test states, scripts, workflows, local execution, reporting, secrets, billing, and administration. Call them directly from any MCP client to build your own pipeline.
+126 tools across authentication, projects, use cases, test cases, test states, scripts, workflows, local execution, reporting, secrets, billing, and administration. Call them directly from any MCP client to build your own pipeline.
 
 **Authentication** — `muggle-remote-auth-status`, `-login`, `-poll`, `-logout`, plus `-api-key-create`, `-list`, `-get`, `-revoke`.
 
-**Projects** — `muggle-remote-project-create`, `-list`, `-get`, `-update`, `-delete`, plus per-project rollups: `-test-results-summary-get`, `-test-runs-summary-get`, `-test-scripts-summary-get`.
+**Projects** — `muggle-remote-project-create`, `-list`, `-get`, `-update`, `-delete`, plus per-project rollups: `-test-results-summary-get`, `-test-runs-summary-get`, `-test-scripts-summary-get`. Pass `primaryResolution`, `compatibleResolutions` and `runStrategy` (`primaryOnly`, the default, `primaryFirst` or `allAtOnce`) to `-create` or `-update` to verify every test case at more than one viewport; `muggle-remote-project-display-resolutions-get` reads a lane's primary, compatible and enabled resolutions and its run strategy.
 
 | Use cases | Purpose |
 | :-------- | :------ |

package/dist/{chunk-XUANXZ2O.js → chunk-7J7DS4SA.js}

@@ -4703,7 +4703,7 @@ function resolveConnectionString(explicit) {
     return explicit;
   return process.env[ENV_CONNECTION_STRING] ?? process.env[ENV_VITE_CONNECTION_STRING] ?? void 0;
 }
-function buildCommonProperties(serviceName, surface) {
+function buildCommonProperties(serviceName, surface, environment) {
   const manifest = readReleaseManifest2();
   const props = {
     source: "client",
@@ -4720,6 +4720,8 @@ function buildCommonProperties(serviceName, surface) {
   const userId = readUserId();
   if (userId !== void 0)
     props.userId = userId;
+  if (environment !== void 0)
+    props.environment = environment;
   return props;
 }
 function initTelemetry(options) {
@@ -4743,7 +4745,7 @@ function initTelemetry(options) {
     client.context.tags[client.context.keys.applicationVersion] = manifest.release;
     client.commonProperties = {
       ...client.commonProperties,
-      ...buildCommonProperties(options.serviceName, options.surface)
+      ...buildCommonProperties(options.serviceName, options.surface, options.environment)
     };
     AppInsights.start();
     state = {
@@ -4862,6 +4864,12 @@ var Surface;
   Surface2["Skill"] = "skill";
   Surface2["Electron"] = "electron";
 })(Surface || (Surface = {}));
+var TelemetryEnvironment;
+(function(TelemetryEnvironment2) {
+  TelemetryEnvironment2["Production"] = "production";
+  TelemetryEnvironment2["Staging"] = "staging";
+  TelemetryEnvironment2["Development"] = "development";
+})(TelemetryEnvironment || (TelemetryEnvironment = {}));
 var ServiceName;
 (function(ServiceName2) {
   ServiceName2["MuggleMcp"] = "muggle-mcp";
@@ -6388,6 +6396,55 @@ var FanOutChoiceInputSchema = z.enum(["followPolicy" /* FollowPolicy */, "primar
   "Optional: followPolicy (default) runs each test case's runSettings records per its lane's fan-out policy; primaryOnly runs only the primary record; allNow runs every record at once"
 );
 
+// packages/mcps/src/mcp/e2e/contracts/run-strategy-types.ts
+var RunStrategy = /* @__PURE__ */ ((RunStrategy2) => {
+  RunStrategy2["PrimaryOnly"] = "primaryOnly";
+  RunStrategy2["PrimaryFirst"] = "primaryFirst";
+  RunStrategy2["AllAtOnce"] = "allAtOnce";
+  return RunStrategy2;
+})(RunStrategy || {});
+
+// packages/mcps/src/mcp/e2e/contracts/project-display-resolution-schemas.ts
+var ProjectResolutionSettingsInputSchema = z.object({
+  primaryResolution: z.nativeEnum(DisplayResolution).optional().describe(
+    'Primary resolution: the viewport every test case in the project is generated and run at first, e.g. "R_1920x1080". Passed without compatibleResolutions, it also clears the compatible resolutions. Omit both to keep the current resolutions; a new project defaults to R_1920x1080 alone.'
+  ),
+  compatibleResolutions: z.array(z.nativeEnum(DisplayResolution)).optional().describe(
+    'Compatible resolutions: further viewports every test case also runs at, one run each, as runStrategy says, e.g. ["R_0390x0844"]. Replaces the whole list; [] clears it. Requires primaryResolution (read the current one with muggle-remote-project-display-resolutions-get) and must not include it; duplicates are dropped.'
+  ),
+  runStrategy: z.nativeEnum(RunStrategy).optional().describe(
+    "Run strategy across the resolutions: primaryOnly runs the primary alone and reports the compatible resolutions as not run; primaryFirst runs the compatible resolutions only after the primary passes; allAtOnce runs every resolution at once. Omit to keep the current one; a project that never set one runs primaryOnly."
+  )
+});
+var ProjectDisplayResolutionsGetInputSchema = z.object({
+  projectId: MuggleEntityIdSchema.describe("Project ID (UUID)"),
+  runEnvironmentType: RunEnvironmentSchema.optional().describe(
+    "Lane to read: 'remote' (cloud runs, the default) or 'local' (developer localhost)."
+  )
+});
+
+// packages/mcps/src/mcp/e2e/contracts/project-resolution-settings-validation.ts
+function addProjectResolutionSettingsIssues(settings, refinementContext) {
+  if (settings.compatibleResolutions === void 0) {
+    return;
+  }
+  if (settings.primaryResolution === void 0) {
+    refinementContext.addIssue({
+      code: "custom",
+      path: ["compatibleResolutions"],
+      message: "compatibleResolutions requires primaryResolution: the project's resolutions are saved as one list, primary first. Pass the current primary (read it with muggle-remote-project-display-resolutions-get) or a new one."
+    });
+    return;
+  }
+  if (settings.compatibleResolutions.includes(settings.primaryResolution)) {
+    refinementContext.addIssue({
+      code: "custom",
+      path: ["compatibleResolutions"],
+      message: `compatibleResolutions must not include the primary resolution ${settings.primaryResolution}; it already runs as the primary.`
+    });
+  }
+}
+
 // packages/mcps/src/mcp/e2e/contracts/secret-group-types.ts
 var SecretTypeEnum = /* @__PURE__ */ ((SecretTypeEnum2) => {
   SecretTypeEnum2["LOGIN_CREDENTIAL"] = "login_credential";
@@ -6574,7 +6631,7 @@ var ProjectCreateInputSchema = z.object({
   organizationId: IdSchema.optional().describe(
     "Organization (UUID) that owns the project; its runs are billed to that organization. Omit for a personal project. The caller must be an active member: list memberships with muggle-remote-organization-list."
   )
-});
+}).extend(ProjectResolutionSettingsInputSchema.shape).superRefine(addProjectResolutionSettingsIssues);
 var ProjectGetInputSchema = z.object({
   projectId: IdSchema.describe("Project ID (UUID) to retrieve")
 });
@@ -6589,7 +6646,7 @@ var ProjectUpdateInputSchema = z.object({
   organizationId: IdSchema.optional().describe(
     "Move a personal project into this organization (UUID). One-way: an organization's project cannot move to another organization or back to personal. The caller must be an active member."
   )
-});
+}).extend(ProjectResolutionSettingsInputSchema.shape).superRefine(addProjectResolutionSettingsIssues);
 var ProjectListInputSchema = PaginationInputSchema.extend({});
 var PrdFileUploadInputSchema = z.object({
   projectId: IdSchema.describe("Project ID (UUID) to associate the PRD file with"),
@@ -7517,6 +7574,57 @@ function mapOrganizationList(response) {
   };
 }
 
+// packages/mcps/src/mcp/tools/e2e/project-display-resolutions-constants.ts
+var KNOWN_DISPLAY_RESOLUTIONS = new Set(Object.values(DisplayResolution));
+var KNOWN_RUN_STRATEGIES = new Set(Object.values(RunStrategy));
+var DEFAULT_RUN_STRATEGY = "primaryOnly" /* PrimaryOnly */;
+
+// packages/mcps/src/mcp/tools/e2e/project-display-resolutions-transform.ts
+function toProjectResolutionUpstreamFields(settings) {
+  const upstreamFields = {};
+  if (settings.primaryResolution !== void 0) {
+    upstreamFields.displayOptions = [
+      .../* @__PURE__ */ new Set([settings.primaryResolution, ...settings.compatibleResolutions ?? []])
+    ];
+  }
+  if (settings.runStrategy !== void 0) {
+    upstreamFields.fanOutPolicy = settings.runStrategy;
+  }
+  return upstreamFields;
+}
+function mapProjectDisplayResolutions(response, input) {
+  const toolInput = input;
+  const envelope = response.data;
+  const displayResolutionAxis = envelope?.effectivePerAxis?.displayResolution ?? envelope?.projectTestSpec?.perAxis?.displayResolution;
+  if (!displayResolutionAxis) {
+    throw new GatewayError({
+      code: "UPSTREAM_ERROR" /* UPSTREAM_ERROR */,
+      message: "project test spec response carried no display resolutions: neither effectivePerAxis.displayResolution nor projectTestSpec.perAxis.displayResolution.",
+      statusCode: response.statusCode
+    });
+  }
+  const runEnvironmentType = toolInput.runEnvironmentType ?? "remote" /* Remote */;
+  const runStrategy = toRunStrategy(envelope?.fanOutPolicy ?? envelope?.projectTestSpec?.fanOutPolicy);
+  const [primaryOptionId, ...compatibleOptionIds] = displayResolutionAxis.default ?? [];
+  return {
+    projectId: toolInput.projectId,
+    runEnvironmentType,
+    primaryResolution: isKnownDisplayResolution(primaryOptionId) ? primaryOptionId : null,
+    compatibleResolutions: toKnownDisplayResolutions(compatibleOptionIds),
+    enabledResolutions: toKnownDisplayResolutions(displayResolutionAxis.enabledValues),
+    runStrategy
+  };
+}
+function isKnownDisplayResolution(optionId) {
+  return optionId !== void 0 && KNOWN_DISPLAY_RESOLUTIONS.has(optionId);
+}
+function toKnownDisplayResolutions(optionIds) {
+  return (optionIds ?? []).filter(isKnownDisplayResolution);
+}
+function toRunStrategy(fanOutPolicy) {
+  return fanOutPolicy !== void 0 && KNOWN_RUN_STRATEGIES.has(fanOutPolicy) ? fanOutPolicy : DEFAULT_RUN_STRATEGY;
+}
+
 // packages/mcps/src/mcp/tools/e2e/run-settings-list-transform.ts
 function mapRunSettingsList(response) {
   const envelope = response.data;
@@ -7726,7 +7834,7 @@ var getWorkflowTimeoutMs = () => getConfig().e2e.workflowTimeoutMs;
 var projectTools = [
   {
     name: "muggle-remote-project-create",
-    description: "Create an E2E acceptance testing project to organize browser tests for a web app. A project groups test scenarios (use cases), specific test steps (test cases), and replayable browser scripts (test scripts) for one application. Create a project first before generating or running any E2E tests. Pass organizationId to create it under an organization the caller belongs to; omit it for a personal project.",
+    description: "Create an E2E acceptance testing project to organize browser tests for a web app. A project groups test scenarios (use cases), specific test steps (test cases), and replayable browser scripts (test scripts) for one application. Create a project first before generating or running any E2E tests. Pass organizationId to create it under an organization the caller belongs to; omit it for a personal project. Pass primaryResolution, compatibleResolutions and runStrategy to verify every test case at more than one viewport (e.g. desktop and phone); omit them for R_1920x1080 alone under primaryOnly.",
     inputSchema: ProjectCreateInputSchema,
     dashboardLink: linkToCreatedProject,
     mapToUpstream: (input) => {
@@ -7734,7 +7842,8 @@ var projectTools = [
       const body = {
         name: toolInput.projectName,
         description: toolInput.description,
-        url: toolInput.url
+        url: toolInput.url,
+        ...toProjectResolutionUpstreamFields(toolInput)
       };
       if (toolInput.organizationId !== void 0) body.organizationId = toolInput.organizationId;
       return {
@@ -7746,7 +7855,7 @@ var projectTools = [
   },
   {
     name: "muggle-remote-project-get",
-    description: "Get details of a specific project by ID.",
+    description: "Get details of a specific project by ID. Its displayOptions is the resolution list last saved on the project, primary first; read the primary and compatible resolutions in effect, and the run strategy, with muggle-remote-project-display-resolutions-get.",
     inputSchema: ProjectGetInputSchema,
     dashboardLink: linkToInputProjectPage("project" /* Project */),
     mapToUpstream: (input) => {
@@ -7759,11 +7868,14 @@ var projectTools = [
   },
   {
     name: "muggle-remote-project-update",
-    description: "Update an existing project's details, or move a personal project into an organization the caller belongs to.",
+    description: "Update an existing project's details, or move a personal project into an organization the caller belongs to. primaryResolution, compatibleResolutions and runStrategy change, on both lanes, the resolutions every test case is verified at and how its runs across them start. The resolutions save as one list: compatibleResolutions requires primaryResolution and replaces the whole compatible list, and primaryResolution alone clears it, so read the current ones with muggle-remote-project-display-resolutions-get and pass back those to keep. runStrategy alone leaves the resolutions unchanged.",
     inputSchema: ProjectUpdateInputSchema,
     mapToUpstream: (input) => {
       const toolInput = input;
-      const body = { id: toolInput.projectId };
+      const body = {
+        id: toolInput.projectId,
+        ...toProjectResolutionUpstreamFields(toolInput)
+      };
       if (toolInput.projectName !== void 0) body.name = toolInput.projectName;
       if (toolInput.description !== void 0) body.description = toolInput.description;
       if (toolInput.url !== void 0) body.url = toolInput.url;
@@ -7775,6 +7887,20 @@ var projectTools = [
       };
     }
   },
+  {
+    name: "muggle-remote-project-display-resolutions-get",
+    description: "Get a project lane's resolutions and run strategy: the resolutions every test case that picks none of its own is verified at, one run each, and how those runs start. Response shape: { projectId, runEnvironmentType, primaryResolution, compatibleResolutions, enabledResolutions, runStrategy }. runStrategy is primaryOnly (the primary runs alone; compatible resolutions are reported as not run), primaryFirst (compatible resolutions run once the primary passes) or allAtOnce (every resolution runs at once); a lane that never set one reports primaryOnly. enabledResolutions are those a test case on the lane may pick. Change them with primaryResolution, compatibleResolutions and runStrategy on muggle-remote-project-update.",
+    inputSchema: ProjectDisplayResolutionsGetInputSchema,
+    mapToUpstream: (input) => {
+      const toolInput = input;
+      return {
+        method: "GET",
+        path: `${MUGGLE_TEST_PREFIX}/projects/${toolInput.projectId}/project-test-spec`,
+        queryParams: { type: toolInput.runEnvironmentType ?? "remote" /* Remote */ }
+      };
+    },
+    mapFromUpstream: mapProjectDisplayResolutions
+  },
   {
     name: "muggle-remote-project-list",
     description: "List projects accessible to the authenticated user: the projects they own plus the projects shared with them. Returns up to 10 items per page by default (max 100). Response includes pagination metadata (totalCount, totalPages, hasMore) \u2014 check `hasMore` to decide whether to fetch additional pages.",
@@ -8058,7 +8184,7 @@ var testCaseTools = [
   },
   {
     name: "muggle-remote-fan-out-plan",
-    description: "Plan a test case's fan-out across its runSettings records in one lane, for a caller that runs the records itself, such as a local run; it starts nothing and writes nothing. Response shape: { testCaseId, laneType, fanOutPolicy, fanOutChoice, isPrimaryKnownGood, isSecondaryGatedOnPrimary, records: [{ runSettingsId, resolutionLabel, displayResolution, isPrimary, testScriptId }], notRunRunSettingsIds }. `records` is in start order, primary first: run records[0] first. When isSecondaryGatedOnPrimary is true, run the rest only if that run passed, and otherwise report each as skipped because the primary failed; when false, run them all. A record's testScriptId is its script on the lane, absent when it has none: replay that script, or generate with the record's runSettingsId and displayResolution. Empty records means the lane has no records for the test case. notRunRunSettingsIds lists the records primaryOnly leaves unrun. Omit fanOutChoice to follow the lane's policy.",
+    description: "Plan a test case's fan-out across its runSettings records in one lane, for a caller that runs the records itself, such as a local run; it starts nothing and writes nothing. Response shape: { testCaseId, laneType, fanOutPolicy, fanOutChoice, isPrimaryKnownGood, isSecondaryGatedOnPrimary, records: [{ runSettingsId, resolutionLabel, displayResolution, isPrimary, testScriptId }], notRunRunSettingsIds }. `records` is in start order, primary first: run records[0] first. When isSecondaryGatedOnPrimary is true, run the rest only if that run passed, and otherwise report each as skipped because the primary failed; when false, run them all. A record's testScriptId is its script on the lane, absent when it has none: replay that script, or generate with the record's runSettingsId and displayResolution. Empty records means the lane has no records for the test case. notRunRunSettingsIds lists the records left unrun because fanOutChoice or the lane's fanOutPolicy is primaryOnly. Omit fanOutChoice to follow the lane's policy.",
     inputSchema: FanOutPlanInputSchema,
     mapToUpstream: (input) => {
       const toolInput = input;
@@ -10922,8 +11048,10 @@ __export(e2e_exports2, {
   PrdFileUploadInputSchema: () => PrdFileUploadInputSchema,
   ProjectCreateInputSchema: () => ProjectCreateInputSchema,
   ProjectDeleteInputSchema: () => ProjectDeleteInputSchema,
+  ProjectDisplayResolutionsGetInputSchema: () => ProjectDisplayResolutionsGetInputSchema,
   ProjectGetInputSchema: () => ProjectGetInputSchema,
   ProjectListInputSchema: () => ProjectListInputSchema,
+  ProjectResolutionSettingsInputSchema: () => ProjectResolutionSettingsInputSchema,
   ProjectTestResultsSummaryInputSchema: () => ProjectTestResultsSummaryInputSchema,
   ProjectTestRunsSummaryInputSchema: () => ProjectTestRunsSummaryInputSchema,
   ProjectTestScriptsSummaryInputSchema: () => ProjectTestScriptsSummaryInputSchema,
@@ -10940,6 +11068,7 @@ __export(e2e_exports2, {
   RunEnvironmentInputSchema: () => RunEnvironmentInputSchema,
   RunEnvironmentSchema: () => RunEnvironmentSchema,
   RunSettingsListInputSchema: () => RunSettingsListInputSchema,
+  RunStrategy: () => RunStrategy,
   SecretCreateInputSchema: () => SecretCreateInputSchema,
   SecretDeleteInputSchema: () => SecretDeleteInputSchema,
   SecretFieldEnum: () => SecretFieldEnum,
@@ -11586,4 +11715,4 @@ function applyLlmSourceSelection(selection, dataDirOverride2) {
   return resolveLlmSourceStatus(dataDirOverride2);
 }
 
-export { ACTION_SCRIPT_FILE_NAME, BASE_REMOTE_NAME, BLOCKING_FRESHNESS_VERDICTS, BRANCH_CREATION_REFLOG_PREFIX, BUSY_EXIT_CODE, CLAUDE_CODE_AUTH_STATUS_ARGS, CLAUDE_CODE_CHECK_TIMEOUT_MS, CLAUDE_CODE_EXECUTABLE, CLAUDE_CODE_EXECUTABLE_NOT_FOUND_CODE, CLAUDE_CODE_SESSION_LINK_ENV_VARS, CLAUDE_CODE_VERSION_ARGS, CheckoutFreshnessVerdict, ClaudeCodeReadiness, DEFAULT_LOCAL_LLM_BASE_URL, DEFAULT_PREFERENCES, ELECTRON_APP_EXECUTABLE_PATH_SEGMENTS, ElectronAppPlatform, ElectronAppReleaseStream, EventName, HTML_VIEW_OFFER_LINE, LLM_CREDENTIALS_FILE_MODE, LLM_CREDENTIALS_FILE_NAME, LLM_SOURCE_HEADER, LLM_SOURCE_OPTIONS, LLM_SOURCE_OWNED_ENV_NAMES, LLM_SOURCE_PROMPT, LLM_SOURCE_SCOPE_NOTE, LlmSource, MASKED_API_KEY_PREFIX, MASKED_API_KEY_VISIBLE_CHARS, MAX_LISTED_INHERITED_COMMITS, ONBOARDING_BLANKET_OPTIONS, ONBOARDING_CHOICE_GROUPS, ONBOARDING_MAX_OFFERS, ONBOARDING_PRIMER_BULLETS, ONBOARDING_PRIMER_HEADLINE, ONBOARDING_TOGGLE_GROUPS, ONBOARDING_TOGGLE_PROMPT, ONBOARDING_UNSELECTED_TOGGLE_VALUE, OnboardingBlanketChoice, OnboardingGroupKind, PREFERENCES_FILE_NAME, PREFERENCES_PROJECT_DIR_NAME, PREFERENCES_SCHEMA, PREFERENCES_VERSION, PREFERENCE_ALLOWED_VALUES, PreferenceKey, PreferenceValue, ProjectPreferencesReconcileOutcome, RUN_SESSIONS_DIR_NAME, RUN_VIEW_DASHBOARD_LINK_LABEL, RunViewError, RuntimeTarget, SCREENSHOT_DIR_RELATIVE_PATH, SCREENSHOT_FILE_EXTENSION, ServiceName, StackCheckVerdict, StackClaimVerdict, StackLeaseEndReason, StackMismatchKind, StackPortAction, Surface, WATCHER_LIFETIME_SECONDS, WATCHER_LIFETIME_UNBOUNDED_SECONDS, __export, __require, applyLlmSourceSelection, applyOnboardingAnswers, assertDeviceCodeClientProvisioned, buildElectronAppChecksumsUrl, buildElectronAppDirName, buildElectronAppReleaseAssetUrl, buildElectronAppReleaseTag, buildLlmSourcePlan, buildOnboardingPlan, buildToolInputJsonSchema, calculateFileChecksum, checkServiceStack, claimServiceStack, compareQueuePlaces, completeOnboarding, createApiKeyWithToken, createChildLogger, decideStackClaim, decideStackReap, decideStackRelease, decideStackRunEnd, decideStackRunStart, deleteApiKeyData, deleteCredentials, describeLlmSourceStatus, describeStackMismatch, e2e_exports2 as e2e_exports, evaluateCheckoutFreshness, findVacatedNeeds, finishStackRun, formatPreferencesOneLiner, getActiveElectronAppReleaseStream, getActiveRuntimeTarget, getApiKey, getApiKeyFilePath, getAuthService, getBundledElectronAppVersion, getCallerCredentials, getCallerCredentialsAsync, getChecksumForPlatform, getConfig, getCredentialsFilePath, getDashboardBaseUrl, getDataDir, getDataDir2, getDisclosureCopy, getDownloadBaseUrl, getElectronAppChecksums, getElectronAppDir, getElectronAppReleaseTagPrefix, getElectronAppSignedFromVersion, getElectronAppVersion, getElectronAppVersionSource, getLlmSourceOption, getLocalQaTools, getLogger, getOnboardingToggleKeys, getPlatformKey, getQaTools, getReleaseSignerIdentityUri, getRunSessionsDir, getValidApiKeyData, getValidCredentials, hasApiKey, hasShownDisclosure, initTelemetry, isElectronAppInstalled, isElectronAppPlatform, isFirstRun, isValidLocalBaseUrl, loadApiKeyData, loadCredentials, local_exports2 as local_exports, lookUpCheckoutPullRequest, markDisclosureShown, markOnboardingCompleted, maskLlmApiKey, mcp_exports, needsConflict, needsOnboarding, openBrowserUrl, parseElectronAppDirName, performLogin, performLogout, pollDeviceCode, portOfUrl, readPreferencesMetadata, readRunView, readServiceStackStatus, reapServiceStack, reconcileProjectPreferences, recordOnboardingSkip, releaseServiceStack, renderRunViewHtml, renderRunViewText, resetConfig, resetLogger, resetPreference, resolveActiveProfile, resolveActiveReleaseStream, resolveActiveReleaseTagPrefix, resolveBlanketPreferences, resolveElectronAppExecutablePath, resolveElectronAppPathOrNull, resolveLlmSourceStatus, resolveOnboardingAnswers, resolvePreferences, resolveRunSessionPath, resolveRuntimeTarget, resolveTogglePreferences, runGitCommand, saveApiKey, saveApiKeyData, saveCredentials, src_exports, startDeviceCodeFlow, startStackRun, stopVacatedServices, toDisplayRef, toolRequiresAuth, track, validatePreference, verifyFileChecksum, writeLlmEnvBlock, writePreferences, writePreferencesWithMetadata, writeRunViewHtml };
+export { ACTION_SCRIPT_FILE_NAME, BASE_REMOTE_NAME, BLOCKING_FRESHNESS_VERDICTS, BRANCH_CREATION_REFLOG_PREFIX, BUSY_EXIT_CODE, CLAUDE_CODE_AUTH_STATUS_ARGS, CLAUDE_CODE_CHECK_TIMEOUT_MS, CLAUDE_CODE_EXECUTABLE, CLAUDE_CODE_EXECUTABLE_NOT_FOUND_CODE, CLAUDE_CODE_SESSION_LINK_ENV_VARS, CLAUDE_CODE_VERSION_ARGS, CheckoutFreshnessVerdict, ClaudeCodeReadiness, DEFAULT_LOCAL_LLM_BASE_URL, DEFAULT_PREFERENCES, ELECTRON_APP_EXECUTABLE_PATH_SEGMENTS, ElectronAppPlatform, ElectronAppReleaseStream, EventName, HTML_VIEW_OFFER_LINE, LLM_CREDENTIALS_FILE_MODE, LLM_CREDENTIALS_FILE_NAME, LLM_SOURCE_HEADER, LLM_SOURCE_OPTIONS, LLM_SOURCE_OWNED_ENV_NAMES, LLM_SOURCE_PROMPT, LLM_SOURCE_SCOPE_NOTE, LlmSource, MASKED_API_KEY_PREFIX, MASKED_API_KEY_VISIBLE_CHARS, MAX_LISTED_INHERITED_COMMITS, ONBOARDING_BLANKET_OPTIONS, ONBOARDING_CHOICE_GROUPS, ONBOARDING_MAX_OFFERS, ONBOARDING_PRIMER_BULLETS, ONBOARDING_PRIMER_HEADLINE, ONBOARDING_TOGGLE_GROUPS, ONBOARDING_TOGGLE_PROMPT, ONBOARDING_UNSELECTED_TOGGLE_VALUE, OnboardingBlanketChoice, OnboardingGroupKind, PREFERENCES_FILE_NAME, PREFERENCES_PROJECT_DIR_NAME, PREFERENCES_SCHEMA, PREFERENCES_VERSION, PREFERENCE_ALLOWED_VALUES, PreferenceKey, PreferenceValue, ProjectPreferencesReconcileOutcome, RUN_SESSIONS_DIR_NAME, RUN_VIEW_DASHBOARD_LINK_LABEL, RunViewError, RuntimeTarget, SCREENSHOT_DIR_RELATIVE_PATH, SCREENSHOT_FILE_EXTENSION, ServiceName, StackCheckVerdict, StackClaimVerdict, StackLeaseEndReason, StackMismatchKind, StackPortAction, Surface, TelemetryEnvironment, WATCHER_LIFETIME_SECONDS, WATCHER_LIFETIME_UNBOUNDED_SECONDS, __export, __require, applyLlmSourceSelection, applyOnboardingAnswers, assertDeviceCodeClientProvisioned, buildElectronAppChecksumsUrl, buildElectronAppDirName, buildElectronAppReleaseAssetUrl, buildElectronAppReleaseTag, buildLlmSourcePlan, buildOnboardingPlan, buildToolInputJsonSchema, calculateFileChecksum, checkServiceStack, claimServiceStack, compareQueuePlaces, completeOnboarding, createApiKeyWithToken, createChildLogger, decideStackClaim, decideStackReap, decideStackRelease, decideStackRunEnd, decideStackRunStart, deleteApiKeyData, deleteCredentials, describeLlmSourceStatus, describeStackMismatch, e2e_exports2 as e2e_exports, evaluateCheckoutFreshness, findVacatedNeeds, finishStackRun, formatPreferencesOneLiner, getActiveElectronAppReleaseStream, getActiveRuntimeTarget, getApiKey, getApiKeyFilePath, getAuthService, getBundledElectronAppVersion, getCallerCredentials, getCallerCredentialsAsync, getChecksumForPlatform, getConfig, getCredentialsFilePath, getDashboardBaseUrl, getDataDir, getDataDir2, getDisclosureCopy, getDownloadBaseUrl, getElectronAppChecksums, getElectronAppDir, getElectronAppReleaseTagPrefix, getElectronAppSignedFromVersion, getElectronAppVersion, getElectronAppVersionSource, getLlmSourceOption, getLocalQaTools, getLogger, getOnboardingToggleKeys, getPlatformKey, getQaTools, getReleaseSignerIdentityUri, getRunSessionsDir, getValidApiKeyData, getValidCredentials, hasApiKey, hasShownDisclosure, initTelemetry, isElectronAppInstalled, isElectronAppPlatform, isFirstRun, isValidLocalBaseUrl, loadApiKeyData, loadCredentials, local_exports2 as local_exports, lookUpCheckoutPullRequest, markDisclosureShown, markOnboardingCompleted, maskLlmApiKey, mcp_exports, needsConflict, needsOnboarding, openBrowserUrl, parseElectronAppDirName, performLogin, performLogout, pollDeviceCode, portOfUrl, readPreferencesMetadata, readRunView, readServiceStackStatus, reapServiceStack, reconcileProjectPreferences, recordOnboardingSkip, releaseServiceStack, renderRunViewHtml, renderRunViewText, resetConfig, resetLogger, resetPreference, resolveActiveProfile, resolveActiveReleaseStream, resolveActiveReleaseTagPrefix, resolveBlanketPreferences, resolveElectronAppExecutablePath, resolveElectronAppPathOrNull, resolveLlmSourceStatus, resolveOnboardingAnswers, resolvePreferences, resolveRunSessionPath, resolveRuntimeTarget, resolveTogglePreferences, runGitCommand, saveApiKey, saveApiKeyData, saveCredentials, src_exports, startDeviceCodeFlow, startStackRun, stopVacatedServices, toDisplayRef, toolRequiresAuth, track, validatePreference, verifyFileChecksum, writeLlmEnvBlock, writePreferences, writePreferencesWithMetadata, writeRunViewHtml };

package/dist/{chunk-RQ43IPVN.js → chunk-7JQWRPFI.js}

@@ -1,4 +1,4 @@
-import { __export, getLogger, getConfig, createChildLogger, buildElectronAppReleaseAssetUrl, getActiveRuntimeTarget, getAuthService, hasApiKey, getElectronAppVersion, getElectronAppDir, getPlatformKey, reconcileProjectPreferences, getDataDir2, PREFERENCES_FILE_NAME, isFirstRun, writePreferences, DEFAULT_PREFERENCES, isElectronAppInstalled, getElectronAppChecksums, getChecksumForPlatform, getElectronAppSignedFromVersion, getReleaseSignerIdentityUri, verifyFileChecksum, resolveElectronAppExecutablePath, calculateFileChecksum, initTelemetry, Surface, ServiceName, track, EventName, getQaTools, getLocalQaTools, performLogout, assertDeviceCodeClientProvisioned, performLogin, getActiveElectronAppReleaseStream, buildToolInputJsonSchema, toolRequiresAuth, getCallerCredentials, parseElectronAppDirName, getElectronAppReleaseTagPrefix, hasShownDisclosure, markDisclosureShown, getBundledElectronAppVersion, getElectronAppVersionSource, getCredentialsFilePath, ElectronAppReleaseStream, buildElectronAppChecksumsUrl, __require, getDisclosureCopy, buildOnboardingPlan, applyOnboardingAnswers, needsOnboarding, buildLlmSourcePlan, describeLlmSourceStatus, resolveLlmSourceStatus, readRunView, RunViewError, renderRunViewText, writeRunViewHtml, openBrowserUrl, evaluateCheckoutFreshness, runGitCommand, BLOCKING_FRESHNESS_VERDICTS, claimServiceStack, BUSY_EXIT_CODE, checkServiceStack, releaseServiceStack, readServiceStackStatus, resolvePreferences, OnboardingBlanketChoice, PreferenceKey, LlmSource, applyLlmSourceSelection, describeStackMismatch, getDataDir, DEFAULT_LOCAL_LLM_BASE_URL, isValidLocalBaseUrl, local_exports } from './chunk-XUANXZ2O.js';
+import { __export, getLogger, TelemetryEnvironment, getConfig, createChildLogger, buildElectronAppReleaseAssetUrl, getActiveRuntimeTarget, getAuthService, hasApiKey, getElectronAppVersion, getElectronAppDir, getPlatformKey, reconcileProjectPreferences, getDataDir2, PREFERENCES_FILE_NAME, isFirstRun, writePreferences, DEFAULT_PREFERENCES, isElectronAppInstalled, getElectronAppChecksums, getChecksumForPlatform, getElectronAppSignedFromVersion, getReleaseSignerIdentityUri, verifyFileChecksum, resolveElectronAppExecutablePath, calculateFileChecksum, initTelemetry, Surface, ServiceName, track, EventName, getQaTools, getLocalQaTools, performLogout, assertDeviceCodeClientProvisioned, performLogin, getActiveElectronAppReleaseStream, buildToolInputJsonSchema, toolRequiresAuth, getCallerCredentials, parseElectronAppDirName, getElectronAppReleaseTagPrefix, hasShownDisclosure, markDisclosureShown, getBundledElectronAppVersion, getElectronAppVersionSource, getCredentialsFilePath, ElectronAppReleaseStream, buildElectronAppChecksumsUrl, __require, getDisclosureCopy, buildOnboardingPlan, applyOnboardingAnswers, needsOnboarding, buildLlmSourcePlan, describeLlmSourceStatus, resolveLlmSourceStatus, readRunView, RunViewError, renderRunViewText, writeRunViewHtml, openBrowserUrl, evaluateCheckoutFreshness, runGitCommand, BLOCKING_FRESHNESS_VERDICTS, claimServiceStack, BUSY_EXIT_CODE, checkServiceStack, releaseServiceStack, readServiceStackStatus, resolvePreferences, OnboardingBlanketChoice, PreferenceKey, LlmSource, applyLlmSourceSelection, describeStackMismatch, getDataDir, DEFAULT_LOCAL_LLM_BASE_URL, isValidLocalBaseUrl, local_exports } from './chunk-7J7DS4SA.js';
 import { Server } from '@modelcontextprotocol/sdk/server/index.js';
 import { ListToolsRequestSchema, CallToolRequestSchema, ListResourcesRequestSchema, ReadResourceRequestSchema } from '@modelcontextprotocol/sdk/types.js';
 import { v4 } from 'uuid';
@@ -1165,7 +1165,7 @@ async function resolveGsScreenshotUrls(report, opts) {
     if (gsUrls.length === 0) {
       return report;
     }
-    const mcps = await import('./src-J6AIYTLA.js');
+    const mcps = await import('./src-QA7TWQ2M.js');
     const credentials = await mcps.getCallerCredentialsAsync();
     if (!credentials.bearerToken && !credentials.apiKey) {
       stderrWrite(
@@ -3979,6 +3979,13 @@ function resolveDisclosureCopy() {
   return `${upstream.slice(0, markerIndex)}${DISCLOSURE_OPT_OUT_COPY}`;
 }
 
+// src/cli/telemetry-environment-constants.ts
+var TELEMETRY_ENVIRONMENT_BY_RUNTIME_TARGET = {
+  ["production" /* Production */]: TelemetryEnvironment.Production,
+  ["staging" /* Staging */]: TelemetryEnvironment.Staging,
+  ["dev" /* Dev */]: TelemetryEnvironment.Development
+};
+
 // src/cli/serve.ts
 var APPLICATIONINSIGHTS_CONNECTION_STRING = "InstrumentationKey=0e28df42-b98e-4089-b3b5-6decefb99447;IngestionEndpoint=https://westus3-1.in.applicationinsights.azure.com/;LiveEndpoint=https://westus3.livediagnostics.monitor.azure.com/;ApplicationId=b64a932f-85ec-40cd-a74b-e181a3660f6c";
 function showDisclosureIfNeeded() {
@@ -4005,7 +4012,8 @@ async function serveCommand(options) {
     initTelemetry({
       serviceName: ServiceName.MuggleMcp,
       surface: Surface.McpLocal,
-      connectionString: APPLICATIONINSIGHTS_CONNECTION_STRING
+      connectionString: APPLICATIONINSIGHTS_CONNECTION_STRING,
+      environment: TELEMETRY_ENVIRONMENT_BY_RUNTIME_TARGET[getActiveRuntimeTarget()]
     });
     track({ name: EventName.SystemStartup, props: { serviceName: ServiceName.MuggleMcp } });
   } catch (err) {
@@ -5012,11 +5020,25 @@ function readRegistryBySessionId(registryDir) {
   }
   return entriesBySessionId;
 }
+function isPidAlive(pid) {
+  try {
+    process.kill(pid, 0);
+    return true;
+  } catch (error) {
+    return error.code === "EPERM";
+  }
+}
+function isSessionRunning(runCommand, registryDir, sessionId) {
+  const registryEntry = readRegistryBySessionId(registryDir).get(sessionId);
+  if (registryEntry !== void 0 && isPidAlive(registryEntry.pid)) return true;
+  const sessions = readClaudeSessions(runCommand, registryDir);
+  if (sessions === null) return null;
+  const session = sessions.find((listed) => listed.sessionId === sessionId);
+  return session !== void 0 && hasLiveProcess(session);
+}
 function waitForSessionExit(runCommand, registryDir, sessionId, sleep2 = blockFor) {
   for (let waitedMs = 0; ; waitedMs += SESSION_EXIT_POLL_MS) {
-    const sessions = readClaudeSessions(runCommand, registryDir);
-    const session = sessions?.find((listed) => listed.sessionId === sessionId);
-    if (sessions !== null && (session === void 0 || !hasLiveProcess(session))) return true;
+    if (isSessionRunning(runCommand, registryDir, sessionId) === false) return true;
     if (waitedMs >= SESSION_EXIT_WAIT_MS) return false;
     sleep2(SESSION_EXIT_POLL_MS);
   }
@@ -5340,13 +5362,22 @@ function startServiceLoops(assessments, nowMs, isPollingDisabled, ownerSessionOf
 }
 
 // src/watch/session-restart.ts
-function resumeSession(target, runCommand, waitForExit) {
+function resumeSession(target, runCommand, waitForExit, isSessionRunning2) {
+  const runningSessionId = target.runningSessionId ?? target.sessionId;
   if (target.runningShortId !== void 0) {
     runCommand("claude", ["stop", target.runningShortId]);
-    if (!waitForExit(target.runningSessionId ?? target.sessionId)) return false;
-  }
-  const resumed = runCommand("claude", ["--resume", target.sessionId, "--bg", target.prompt], { cwd: target.cwd });
-  return resumed?.exitCode === 0;
+    if (!waitForExit(runningSessionId)) return { outcome: "still-running" /* StillRunning */ };
+  } else if (isSessionRunning2(runningSessionId)) {
+    return { outcome: "still-running" /* StillRunning */ };
+  }
+  const resumed = runCommand("claude", ["--resume", target.sessionId, "--bg", target.prompt], { cwd: target.cwd, isStderrCaptured: true });
+  if (resumed?.exitCode !== 0) return { outcome: "failed" /* Failed */ };
+  const backgroundedShortId = resumed.stdout.match(BACKGROUNDED_SESSION_ID_PATTERN)?.[1];
+  const ownShortIds = [shortSessionId(target.sessionId), shortSessionId(runningSessionId), target.runningShortId];
+  if (backgroundedShortId === void 0 || ownShortIds.includes(backgroundedShortId)) return { outcome: "continued" /* Continued */ };
+  const isCopyStopped = runCommand("claude", ["stop", backgroundedShortId])?.exitCode === 0;
+  const isCopyRemoved = isCopyStopped && runCommand("claude", ["rm", backgroundedShortId])?.exitCode === 0;
+  return { outcome: "copied" /* Copied */, copyShortId: backgroundedShortId, isCopyRemoved };
 }
 
 // src/watch/recovery.ts
@@ -5427,20 +5458,31 @@ function executeWatcherRecoveries(planned, attemptsByOwner, deps) {
       });
       return { ...recovery, cwd, outcome: "restart-refused" /* RestartRefused */ };
     }
-    const isResumed = resumeSession(
+    const resumed = resumeSession(
       { sessionId: recovery.ownerSessionId, cwd, prompt, runningShortId: recovery.ownerShortId, runningSessionId: recovery.ownerRunningSessionId },
       deps.runCommand,
-      deps.waitForExit
+      deps.waitForExit,
+      deps.isSessionRunning
     );
-    if (isResumed) markServiceLoopReportDelivered(recovery.slotDir);
-    else countAttempt(recovery.ownerSessionId);
-    events.push({
-      at: nowIso,
-      kind: isResumed ? "resumed" /* Resumed */ : "resume-failed" /* ResumeFailed */,
-      subject,
-      message: `${isResumed ? "woke it" : "could not wake it"} with ${prompt}`
-    });
-    return { ...recovery, cwd, outcome: isResumed ? "resumed" /* Resumed */ : "resume-failed" /* ResumeFailed */ };
+    if (resumed.outcome === "continued" /* Continued */) {
+      markServiceLoopReportDelivered(recovery.slotDir);
+      events.push({ at: nowIso, kind: "resumed" /* Resumed */, subject, message: `woke it with ${prompt}` });
+      return { ...recovery, cwd, outcome: "resumed" /* Resumed */ };
+    }
+    countAttempt(recovery.ownerSessionId);
+    if (resumed.outcome === "copied" /* Copied */) {
+      const copyNote = resumed.isCopyRemoved ? "stopped and removed the copy" : `could not remove the copy, run claude rm ${resumed.copyShortId}`;
+      events.push({
+        at: nowIso,
+        kind: "resume-forked" /* ResumeForked */,
+        subject,
+        message: `could not wake it: Claude still ran it and started copy ${resumed.copyShortId} instead; ${copyNote}`
+      });
+      return { ...recovery, cwd, outcome: "resume-forked" /* ResumeForked */ };
+    }
+    const reason = resumed.outcome === "still-running" /* StillRunning */ ? "it is still running, so a resume would start a copy" : `${prompt} did not start`;
+    events.push({ at: nowIso, kind: "resume-failed" /* ResumeFailed */, subject, message: `could not wake it: ${reason}` });
+    return { ...recovery, cwd, outcome: "resume-failed" /* ResumeFailed */ };
   });
   return { recoveries, attemptsByOwner: nextAttemptsByOwner, events };
 }
@@ -6440,7 +6482,7 @@ function idleSinceBySessionId(runningSessions) {
   );
 }
 function isRestartAttempted(recovery) {
-  return recovery.outcome === "resumed" /* Resumed */ || recovery.outcome === "resume-failed" /* ResumeFailed */;
+  return recovery.outcome === "resumed" /* Resumed */ || recovery.outcome === "resume-failed" /* ResumeFailed */ || recovery.outcome === "resume-forked" /* ResumeForked */;
 }
 function watchOwnersOf(slots) {
   return new Set(slots.flatMap((slot) => !isTerminalPrState(slot.live?.state) && slot.ownerSessionId ? [slot.ownerSessionId] : []));
@@ -6484,6 +6526,9 @@ function canRestartVia(deps) {
 function waitForExitVia(deps) {
   return (sessionId) => waitForSessionExit(deps.runCommand, deps.paths.claudeSessionRegistryDir, sessionId, deps.sleep);
 }
+function isSessionRunningVia(deps) {
+  return (sessionId) => isSessionRunning(deps.runCommand, deps.paths.claudeSessionRegistryDir, sessionId) !== false;
+}
 function reloadTargetAtMs(deps, installs) {
   return resolveReloadTargetAtMs(installs, readMugglePluginUpdatedAtMs(deps.paths.claudeInstalledPluginsFile));
 }
@@ -6551,7 +6596,8 @@ function runTickUnlocked(deps) {
       claudeProjectsDir: deps.paths.claudeProjectsDir,
       nowMs: deps.nowMs,
       canRestartIn: canRestartVia(deps),
-      waitForExit: waitForExitVia(deps)
+      waitForExit: waitForExitVia(deps),
+      isSessionRunning: isSessionRunningVia(deps)
     });
     recoveries = executed.recoveries;
     recoveryAttemptsByOwner = executed.attemptsByOwner;

package/dist/cli.js

@@ -1,6 +1,6 @@
 #!/usr/bin/env node
-import { runCli } from './chunk-RQ43IPVN.js';
-import './chunk-XUANXZ2O.js';
+import { runCli } from './chunk-7JQWRPFI.js';
+import './chunk-7J7DS4SA.js';
 
 // src/cli/main.ts
 runCli().catch((error) => {

package/dist/index.js

@@ -1,2 +1,2 @@
-export { src_exports as commands, createUnifiedMcpServer, server_exports as server } from './chunk-RQ43IPVN.js';
-export { createChildLogger, e2e_exports as e2e, getConfig, getLocalQaTools, getLogger, getQaTools, local_exports as localQa, mcp_exports as mcp, e2e_exports as qa, src_exports as shared } from './chunk-XUANXZ2O.js';
+export { src_exports as commands, createUnifiedMcpServer, server_exports as server } from './chunk-7JQWRPFI.js';
+export { createChildLogger, e2e_exports as e2e, getConfig, getLocalQaTools, getLogger, getQaTools, local_exports as localQa, mcp_exports as mcp, e2e_exports as qa, src_exports as shared } from './chunk-7J7DS4SA.js';

package/dist/plugin/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "muggle",
   "description": "Real-browser E2E acceptance for your coding agent: test the flows you changed and open the PR with screenshots. Built for flows other tools can't reach: managed test identities with live inboxes (magic links, email OTP and 2FA, verification mail, password resets), an encrypted secret store for login and test credentials, and a test graph that runs prerequisite flows first. Plain-English tests become replayable scripts on localhost or preview URLs.",
-  "version": "5.31.0-staging.271",
+  "version": "5.31.0-staging.273",
   "author": {
     "name": "Muggle AI",
     "email": "support@muggle-ai.com"

package/dist/plugin/.codex-mcp.json

@@ -4,7 +4,7 @@
       "command": "npx",
       "args": [
         "-y",
-        "@muggleai/works@5.31.0-staging.271",
+        "@muggleai/works@5.31.0-staging.273",
         "serve"
       ],
       "env": {

package/dist/plugin/.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "muggle",
-  "version": "5.31.0-staging.271",
+  "version": "5.31.0-staging.273",
   "description": "Real-browser E2E acceptance for your coding agent: test the flows you changed and open the PR with screenshots. Built for flows other tools can't reach: managed test identities with live inboxes (magic links, email OTP and 2FA, verification mail, password resets), an encrypted secret store for login and test credentials, and a test graph that runs prerequisite flows first. Plain-English tests become replayable scripts on localhost or preview URLs.",
   "author": {
     "name": "Muggle AI",

package/dist/plugin/.cursor-plugin/plugin.json

@@ -2,7 +2,7 @@
   "name": "muggle",
   "displayName": "Muggle AI",
   "description": "Real-browser E2E acceptance for your coding agent: test the flows you changed and open the PR with screenshots. Built for flows other tools can't reach: managed test identities with live inboxes (magic links, email OTP and 2FA, verification mail, password resets), an encrypted secret store for login and test credentials, and a test graph that runs prerequisite flows first. Plain-English tests become replayable scripts on localhost or preview URLs.",
-  "version": "5.31.0-staging.271",
+  "version": "5.31.0-staging.273",
   "author": {
     "name": "Muggle AI",
     "email": "support@muggle-ai.com"

package/dist/plugin/.mcp.json

@@ -4,7 +4,7 @@
       "command": "npx",
       "args": [
         "-y",
-        "@muggleai/works@5.31.0-staging.271",
+        "@muggleai/works@5.31.0-staging.273",
         "serve"
       ]
     }

package/dist/plugin/skills/muggle-pr-followup/state-schemas.md

@@ -61,7 +61,7 @@ The slot's opt-in to recovery by the muggle watch service (`muggle watch`, a per
 
 Recovery requires all of: the slot is registered, the PR is open on GitHub, no loop is heard (no loop process carries the slot's `--slot` path, or the one that does prints into a session that has ended), the heartbeat is older than the re-arm hand-off, `polling.disabled` is absent, `owner.json` names an owner, and that owner is not mid-turn. The service then runs `pr-watch-arm.sh` itself, detached, with `MUGGLE_PR_WATCH_LISTENER=muggle-watch`. The loop records [`watch-listener`](#watch-listener) and writes its output to [`watch-service.out`](#watch-serviceout), and no session spends a turn on it.
 
-When that loop exits on an `EVENT` or `TERMINAL` line, the service wakes the owner with `/muggle:muggle-pr-followup <slug> <n> --wake=<event|terminal>`, naming that slot and PR only. It resumes the owner in the background from the project dir its transcript records. A running owner is woken only as a background session idle between turns: it is stopped first, and only after a throwaway background launch confirms Claude will resume it in that workspace. An owner that is mid-turn, waiting on its user, or open in a terminal keeps the report, and the service alerts once the report is overdue. The service never wakes the ended owner of a merged or closed PR; it closes that slot out instead. An `EXPIRED` loop is restarted without waking anyone. The service caps loop starts and wakes per pass, caps failed wakes per owner, and holds every wake while the account sits at a usage limit.
+When that loop exits on an `EVENT` or `TERMINAL` line, the service wakes the owner with `/muggle:muggle-pr-followup <slug> <n> --wake=<event|terminal>`, naming that slot and PR only. It resumes the owner in the background from the project dir its transcript records. A running owner is woken only as a background session idle between turns: it is stopped first, and only after a throwaway background launch confirms Claude will resume it in that workspace. The resume runs only once neither `claude agents` nor Claude's own session registry shows a process for the owner. If Claude still answers with a copy under a new id, the service stops and removes that copy and counts the wake as failed, so the agent view never gains a duplicate. An owner that is mid-turn, waiting on its user, or open in a terminal keeps the report, and the service alerts once the report is overdue. The service never wakes the ended owner of a merged or closed PR; it closes that slot out instead. An `EXPIRED` loop is restarted without waking anyone. The service caps loop starts and wakes per pass, caps failed wakes per owner, and holds every wake while the account sits at a usage limit.
 
 **Absent ⇒ observed only.** The service still reports on the slot's watcher, CI and branch standing, but never starts its loop or wakes its owner. Slots armed before this file existed stay that way until their next arm. The file lives in the slot, so a `.stopped` rename or a `result.md` retires it with the slot.
 

package/dist/release-manifest.json

@@ -1,7 +1,7 @@
 {
   "release": "5.26.0",
-  "buildId": "run-271-1",
-  "commitSha": "5036fc9c7a6ea65a6269500436e8eb9b6c443413",
-  "buildTime": "2026-11-01T05:29:33Z",
+  "buildId": "run-273-1",
+  "commitSha": "79c091d63782313be3672ba64db5ae65c44e2c2b",
+  "buildTime": "2026-11-01T07:21:17Z",
   "serviceName": "muggle-ai-works-mcp"
 }

package/dist/{src-J6AIYTLA.js → src-QA7TWQ2M.js}

@@ -1 +1 @@
-export { ACTION_SCRIPT_FILE_NAME, BASE_REMOTE_NAME, BLOCKING_FRESHNESS_VERDICTS, BRANCH_CREATION_REFLOG_PREFIX, BUSY_EXIT_CODE, CLAUDE_CODE_AUTH_STATUS_ARGS, CLAUDE_CODE_CHECK_TIMEOUT_MS, CLAUDE_CODE_EXECUTABLE, CLAUDE_CODE_EXECUTABLE_NOT_FOUND_CODE, CLAUDE_CODE_SESSION_LINK_ENV_VARS, CLAUDE_CODE_VERSION_ARGS, CheckoutFreshnessVerdict, ClaudeCodeReadiness, DEFAULT_LOCAL_LLM_BASE_URL, DEFAULT_PREFERENCES, ELECTRON_APP_EXECUTABLE_PATH_SEGMENTS, ElectronAppPlatform, ElectronAppReleaseStream, HTML_VIEW_OFFER_LINE, LLM_CREDENTIALS_FILE_MODE, LLM_CREDENTIALS_FILE_NAME, LLM_SOURCE_HEADER, LLM_SOURCE_OPTIONS, LLM_SOURCE_OWNED_ENV_NAMES, LLM_SOURCE_PROMPT, LLM_SOURCE_SCOPE_NOTE, LlmSource, MASKED_API_KEY_PREFIX, MASKED_API_KEY_VISIBLE_CHARS, MAX_LISTED_INHERITED_COMMITS, ONBOARDING_BLANKET_OPTIONS, ONBOARDING_CHOICE_GROUPS, ONBOARDING_MAX_OFFERS, ONBOARDING_PRIMER_BULLETS, ONBOARDING_PRIMER_HEADLINE, ONBOARDING_TOGGLE_GROUPS, ONBOARDING_TOGGLE_PROMPT, ONBOARDING_UNSELECTED_TOGGLE_VALUE, OnboardingBlanketChoice, OnboardingGroupKind, PREFERENCES_FILE_NAME, PREFERENCES_PROJECT_DIR_NAME, PREFERENCES_SCHEMA, PREFERENCES_VERSION, PREFERENCE_ALLOWED_VALUES, PreferenceKey, PreferenceValue, ProjectPreferencesReconcileOutcome, RUN_SESSIONS_DIR_NAME, RUN_VIEW_DASHBOARD_LINK_LABEL, RunViewError, RuntimeTarget, SCREENSHOT_DIR_RELATIVE_PATH, SCREENSHOT_FILE_EXTENSION, StackCheckVerdict, StackClaimVerdict, StackLeaseEndReason, StackMismatchKind, StackPortAction, WATCHER_LIFETIME_SECONDS, WATCHER_LIFETIME_UNBOUNDED_SECONDS, applyLlmSourceSelection, applyOnboardingAnswers, assertDeviceCodeClientProvisioned, buildElectronAppChecksumsUrl, buildElectronAppDirName, buildElectronAppReleaseAssetUrl, buildElectronAppReleaseTag, buildLlmSourcePlan, buildOnboardingPlan, buildToolInputJsonSchema, calculateFileChecksum, checkServiceStack, claimServiceStack, compareQueuePlaces, completeOnboarding, createApiKeyWithToken, createChildLogger, decideStackClaim, decideStackReap, decideStackRelease, decideStackRunEnd, decideStackRunStart, deleteApiKeyData, deleteCredentials, describeLlmSourceStatus, describeStackMismatch, e2e_exports as e2e, evaluateCheckoutFreshness, findVacatedNeeds, finishStackRun, formatPreferencesOneLiner, getActiveElectronAppReleaseStream, getActiveRuntimeTarget, getApiKey, getApiKeyFilePath, getAuthService, getBundledElectronAppVersion, getCallerCredentials, getCallerCredentialsAsync, getChecksumForPlatform, getConfig, getCredentialsFilePath, getDashboardBaseUrl, getDataDir2 as getDataDir, getDownloadBaseUrl, getElectronAppChecksums, getElectronAppDir, getElectronAppReleaseTagPrefix, getElectronAppSignedFromVersion, getElectronAppVersion, getElectronAppVersionSource, getLlmSourceOption, getLocalQaTools, getLogger, getOnboardingToggleKeys, getPlatformKey, getQaTools, getReleaseSignerIdentityUri, getRunSessionsDir, getValidApiKeyData, getValidCredentials, hasApiKey, isElectronAppInstalled, isElectronAppPlatform, isFirstRun, isValidLocalBaseUrl, loadApiKeyData, loadCredentials, local_exports as localQa, lookUpCheckoutPullRequest, markOnboardingCompleted, maskLlmApiKey, mcp_exports as mcp, needsConflict, needsOnboarding, openBrowserUrl, parseElectronAppDirName, performLogin, performLogout, pollDeviceCode, portOfUrl, e2e_exports as qa, readPreferencesMetadata, readRunView, readServiceStackStatus, reapServiceStack, reconcileProjectPreferences, recordOnboardingSkip, releaseServiceStack, renderRunViewHtml, renderRunViewText, resetConfig, resetLogger, resetPreference, resolveActiveProfile, resolveActiveReleaseStream, resolveActiveReleaseTagPrefix, resolveBlanketPreferences, resolveElectronAppExecutablePath, resolveElectronAppPathOrNull, resolveLlmSourceStatus, resolveOnboardingAnswers, resolvePreferences, resolveRunSessionPath, resolveRuntimeTarget, resolveTogglePreferences, runGitCommand, saveApiKey, saveApiKeyData, saveCredentials, startDeviceCodeFlow, startStackRun, stopVacatedServices, toDisplayRef, toolRequiresAuth, validatePreference, verifyFileChecksum, writeLlmEnvBlock, writePreferences, writePreferencesWithMetadata, writeRunViewHtml } from './chunk-XUANXZ2O.js';
+export { ACTION_SCRIPT_FILE_NAME, BASE_REMOTE_NAME, BLOCKING_FRESHNESS_VERDICTS, BRANCH_CREATION_REFLOG_PREFIX, BUSY_EXIT_CODE, CLAUDE_CODE_AUTH_STATUS_ARGS, CLAUDE_CODE_CHECK_TIMEOUT_MS, CLAUDE_CODE_EXECUTABLE, CLAUDE_CODE_EXECUTABLE_NOT_FOUND_CODE, CLAUDE_CODE_SESSION_LINK_ENV_VARS, CLAUDE_CODE_VERSION_ARGS, CheckoutFreshnessVerdict, ClaudeCodeReadiness, DEFAULT_LOCAL_LLM_BASE_URL, DEFAULT_PREFERENCES, ELECTRON_APP_EXECUTABLE_PATH_SEGMENTS, ElectronAppPlatform, ElectronAppReleaseStream, HTML_VIEW_OFFER_LINE, LLM_CREDENTIALS_FILE_MODE, LLM_CREDENTIALS_FILE_NAME, LLM_SOURCE_HEADER, LLM_SOURCE_OPTIONS, LLM_SOURCE_OWNED_ENV_NAMES, LLM_SOURCE_PROMPT, LLM_SOURCE_SCOPE_NOTE, LlmSource, MASKED_API_KEY_PREFIX, MASKED_API_KEY_VISIBLE_CHARS, MAX_LISTED_INHERITED_COMMITS, ONBOARDING_BLANKET_OPTIONS, ONBOARDING_CHOICE_GROUPS, ONBOARDING_MAX_OFFERS, ONBOARDING_PRIMER_BULLETS, ONBOARDING_PRIMER_HEADLINE, ONBOARDING_TOGGLE_GROUPS, ONBOARDING_TOGGLE_PROMPT, ONBOARDING_UNSELECTED_TOGGLE_VALUE, OnboardingBlanketChoice, OnboardingGroupKind, PREFERENCES_FILE_NAME, PREFERENCES_PROJECT_DIR_NAME, PREFERENCES_SCHEMA, PREFERENCES_VERSION, PREFERENCE_ALLOWED_VALUES, PreferenceKey, PreferenceValue, ProjectPreferencesReconcileOutcome, RUN_SESSIONS_DIR_NAME, RUN_VIEW_DASHBOARD_LINK_LABEL, RunViewError, RuntimeTarget, SCREENSHOT_DIR_RELATIVE_PATH, SCREENSHOT_FILE_EXTENSION, StackCheckVerdict, StackClaimVerdict, StackLeaseEndReason, StackMismatchKind, StackPortAction, WATCHER_LIFETIME_SECONDS, WATCHER_LIFETIME_UNBOUNDED_SECONDS, applyLlmSourceSelection, applyOnboardingAnswers, assertDeviceCodeClientProvisioned, buildElectronAppChecksumsUrl, buildElectronAppDirName, buildElectronAppReleaseAssetUrl, buildElectronAppReleaseTag, buildLlmSourcePlan, buildOnboardingPlan, buildToolInputJsonSchema, calculateFileChecksum, checkServiceStack, claimServiceStack, compareQueuePlaces, completeOnboarding, createApiKeyWithToken, createChildLogger, decideStackClaim, decideStackReap, decideStackRelease, decideStackRunEnd, decideStackRunStart, deleteApiKeyData, deleteCredentials, describeLlmSourceStatus, describeStackMismatch, e2e_exports as e2e, evaluateCheckoutFreshness, findVacatedNeeds, finishStackRun, formatPreferencesOneLiner, getActiveElectronAppReleaseStream, getActiveRuntimeTarget, getApiKey, getApiKeyFilePath, getAuthService, getBundledElectronAppVersion, getCallerCredentials, getCallerCredentialsAsync, getChecksumForPlatform, getConfig, getCredentialsFilePath, getDashboardBaseUrl, getDataDir2 as getDataDir, getDownloadBaseUrl, getElectronAppChecksums, getElectronAppDir, getElectronAppReleaseTagPrefix, getElectronAppSignedFromVersion, getElectronAppVersion, getElectronAppVersionSource, getLlmSourceOption, getLocalQaTools, getLogger, getOnboardingToggleKeys, getPlatformKey, getQaTools, getReleaseSignerIdentityUri, getRunSessionsDir, getValidApiKeyData, getValidCredentials, hasApiKey, isElectronAppInstalled, isElectronAppPlatform, isFirstRun, isValidLocalBaseUrl, loadApiKeyData, loadCredentials, local_exports as localQa, lookUpCheckoutPullRequest, markOnboardingCompleted, maskLlmApiKey, mcp_exports as mcp, needsConflict, needsOnboarding, openBrowserUrl, parseElectronAppDirName, performLogin, performLogout, pollDeviceCode, portOfUrl, e2e_exports as qa, readPreferencesMetadata, readRunView, readServiceStackStatus, reapServiceStack, reconcileProjectPreferences, recordOnboardingSkip, releaseServiceStack, renderRunViewHtml, renderRunViewText, resetConfig, resetLogger, resetPreference, resolveActiveProfile, resolveActiveReleaseStream, resolveActiveReleaseTagPrefix, resolveBlanketPreferences, resolveElectronAppExecutablePath, resolveElectronAppPathOrNull, resolveLlmSourceStatus, resolveOnboardingAnswers, resolvePreferences, resolveRunSessionPath, resolveRuntimeTarget, resolveTogglePreferences, runGitCommand, saveApiKey, saveApiKeyData, saveCredentials, startDeviceCodeFlow, startStackRun, stopVacatedServices, toDisplayRef, toolRequiresAuth, validatePreference, verifyFileChecksum, writeLlmEnvBlock, writePreferences, writePreferencesWithMetadata, writeRunViewHtml } from './chunk-7J7DS4SA.js';

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@muggleai/works",
   "mcpName": "io.github.multiplex-ai/muggle",
-  "version": "5.31.0-staging.271",
+  "version": "5.31.0-staging.273",
   "description": "13 years of Microsoft hyperscale engineering, packaged as a harness for your AI coding agent.",
   "type": "module",
   "main": "dist/index.js",

package/plugin/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "muggle",
   "description": "Real-browser E2E acceptance for your coding agent: test the flows you changed and open the PR with screenshots. Built for flows other tools can't reach: managed test identities with live inboxes (magic links, email OTP and 2FA, verification mail, password resets), an encrypted secret store for login and test credentials, and a test graph that runs prerequisite flows first. Plain-English tests become replayable scripts on localhost or preview URLs.",
-  "version": "5.31.0-staging.271",
+  "version": "5.31.0-staging.273",
   "author": {
     "name": "Muggle AI",
     "email": "support@muggle-ai.com"

package/plugin/.codex-mcp.json

@@ -4,7 +4,7 @@
       "command": "npx",
       "args": [
         "-y",
-        "@muggleai/works@5.31.0-staging.271",
+        "@muggleai/works@5.31.0-staging.273",
         "serve"
       ],
       "env": {

package/plugin/.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "muggle",
-  "version": "5.31.0-staging.271",
+  "version": "5.31.0-staging.273",
   "description": "Real-browser E2E acceptance for your coding agent: test the flows you changed and open the PR with screenshots. Built for flows other tools can't reach: managed test identities with live inboxes (magic links, email OTP and 2FA, verification mail, password resets), an encrypted secret store for login and test credentials, and a test graph that runs prerequisite flows first. Plain-English tests become replayable scripts on localhost or preview URLs.",
   "author": {
     "name": "Muggle AI",

package/plugin/.cursor-plugin/plugin.json

@@ -2,7 +2,7 @@
   "name": "muggle",
   "displayName": "Muggle AI",
   "description": "Real-browser E2E acceptance for your coding agent: test the flows you changed and open the PR with screenshots. Built for flows other tools can't reach: managed test identities with live inboxes (magic links, email OTP and 2FA, verification mail, password resets), an encrypted secret store for login and test credentials, and a test graph that runs prerequisite flows first. Plain-English tests become replayable scripts on localhost or preview URLs.",
-  "version": "5.31.0-staging.271",
+  "version": "5.31.0-staging.273",
   "author": {
     "name": "Muggle AI",
     "email": "support@muggle-ai.com"

package/plugin/.mcp.json

@@ -4,7 +4,7 @@
       "command": "npx",
       "args": [
         "-y",
-        "@muggleai/works@5.31.0-staging.271",
+        "@muggleai/works@5.31.0-staging.273",
         "serve"
       ]
     }

package/plugin/skills/muggle-pr-followup/state-schemas.md

@@ -61,7 +61,7 @@ The slot's opt-in to recovery by the muggle watch service (`muggle watch`, a per
 
 Recovery requires all of: the slot is registered, the PR is open on GitHub, no loop is heard (no loop process carries the slot's `--slot` path, or the one that does prints into a session that has ended), the heartbeat is older than the re-arm hand-off, `polling.disabled` is absent, `owner.json` names an owner, and that owner is not mid-turn. The service then runs `pr-watch-arm.sh` itself, detached, with `MUGGLE_PR_WATCH_LISTENER=muggle-watch`. The loop records [`watch-listener`](#watch-listener) and writes its output to [`watch-service.out`](#watch-serviceout), and no session spends a turn on it.
 
-When that loop exits on an `EVENT` or `TERMINAL` line, the service wakes the owner with `/muggle:muggle-pr-followup <slug> <n> --wake=<event|terminal>`, naming that slot and PR only. It resumes the owner in the background from the project dir its transcript records. A running owner is woken only as a background session idle between turns: it is stopped first, and only after a throwaway background launch confirms Claude will resume it in that workspace. An owner that is mid-turn, waiting on its user, or open in a terminal keeps the report, and the service alerts once the report is overdue. The service never wakes the ended owner of a merged or closed PR; it closes that slot out instead. An `EXPIRED` loop is restarted without waking anyone. The service caps loop starts and wakes per pass, caps failed wakes per owner, and holds every wake while the account sits at a usage limit.
+When that loop exits on an `EVENT` or `TERMINAL` line, the service wakes the owner with `/muggle:muggle-pr-followup <slug> <n> --wake=<event|terminal>`, naming that slot and PR only. It resumes the owner in the background from the project dir its transcript records. A running owner is woken only as a background session idle between turns: it is stopped first, and only after a throwaway background launch confirms Claude will resume it in that workspace. The resume runs only once neither `claude agents` nor Claude's own session registry shows a process for the owner. If Claude still answers with a copy under a new id, the service stops and removes that copy and counts the wake as failed, so the agent view never gains a duplicate. An owner that is mid-turn, waiting on its user, or open in a terminal keeps the report, and the service alerts once the report is overdue. The service never wakes the ended owner of a merged or closed PR; it closes that slot out instead. An `EXPIRED` loop is restarted without waking anyone. The service caps loop starts and wakes per pass, caps failed wakes per owner, and holds every wake while the account sits at a usage limit.
 
 **Absent ⇒ observed only.** The service still reports on the slot's watcher, CI and branch standing, but never starts its loop or wakes its owner. Slots armed before this file existed stay that way until their next arm. The file lives in the slot, so a `.stopped` rename or a `result.md` retires it with the slot.
 

package/scripts/muggle-watch/README.md

@@ -4,7 +4,7 @@ A standing per-user scheduled job that runs `muggle watch tick` every few minute
 
 - **Tracks** active Claude Code sessions (`claude agents --json`) and Codex sessions (recent `~/.codex/sessions` rollouts). It also tracks their processes: claude, codex, muggle MCP servers, and PR watch loops. Each live Claude session is stamped with when it last loaded its plugins (its process start, or its newest muggle MCP server start after a `/reload-plugins`) and the muggle works build installed then, read off `~/.muggle-ai/postinstall.log`.
 - **Checks every open PR-watch slot** under `~/.muggle-ai/muggle-do/sessions/`. Is its loop running, and heard by a live session? Is CI red? Is the branch behind or conflicting? Are review threads unresolved?
-- **Keeps registered watchers running.** A slot carrying `watch-registration.json` (written by `pr-watch-arm.sh` on every arm) whose loop is dead or unheard gets a loop the service starts itself, as a program outside every session. When that loop reports new activity or the PR ending, the owner session is woken with `/muggle:muggle-pr-followup <slug> <n> --wake=<event|terminal>` for that one PR, but only between its turns. The service never addresses a review, pushes, or pushes anything else into a session.
+- **Keeps registered watchers running.** A slot carrying `watch-registration.json` (written by `pr-watch-arm.sh` on every arm) whose loop is dead or unheard gets a loop the service starts itself, as a program outside every session. When that loop reports new activity or the PR ending, the owner session is woken with `/muggle:muggle-pr-followup <slug> <n> --wake=<event|terminal>` for that one PR, but only between its turns. A wake that Claude answers with a copy of the session under a new id counts as failed, and the copy is stopped and removed. The service never addresses a review, pushes, or pushes anything else into a session.
 - **Closes out finished slots.** A slot whose PR GitHub reports merged or closed, with no loop and no running owner, is finalized the way a watcher's terminal tick would: `prs.json` takes the state, `followup.log` the terminal line, and `result.md` is written. Only local files change; post-merge cleanup is left to a session and recorded as not run.
 - **Keeps sessions on the installed build.** A session loaded before the newest global install (or a later user-scope update of the muggle plugin) is behind; other plugins updating restart nothing. Idle background sessions are respawned onto the new build, up to 3 per tick, keeping their own id and name. A session that owns open PR watches is left running. Before a restart, a throwaway background launch in the session's workspace confirms Claude will bring it back; a workspace that refuses is left running and named once. Busy or waiting sessions wait for a later tick, and terminal sessions get one notification to run `/reload-plugins`. No interface runs a slash command inside another session.
 - **Keeps sessions on Remote Control.** With `remoteControlAtStartup` on in Claude's settings, a live background session is off when its `~/.claude/sessions/<pid>.json` entry has no Remote Control link, or its process started before the current claude.ai login took effect, so its link serves another account. Such a session is respawned once its turn is over (registry status `idle` or `shell`), sharing the 3-per-tick cap with the reload. That keeps the same session id, so there is no duplicate in the agent view, and nothing is typed into it. Background commands carry over to the restarted process, PR watch loops included, so they keep reporting to that session. Each restart opens a new Remote Control session, so a session already on is never restarted. A restarted session gets 30 minutes to link, and one still off after 2 restarts in a day is left alone. Terminal sessions that are off are listed for `/remote-control`.