@ak--47/dungeon-master 1.3.1 → 1.4.1

package/dungeons/vertical/social.js

@@ -1,7 +1,7 @@
 // ── TWEAK THESE ──
 const SEED = "harness-social";
-const num_days = 100;
-const num_users = 5_000;
+const num_days = 120;
+const num_users = 10_000;
 const avg_events_per_user_per_day = 1.2;
 let token = "your-mixpanel-token";
 
@@ -107,7 +107,7 @@ const chance = u.initChance(SEED);
  * receive follow-backs and post more frequently.
  *
  * -------------------------------------------------------------------------------------
- * 3. ALGORITHM CHANGE (event)
+ * 3. ALGORITHM CHANGE (everything)
  * -------------------------------------------------------------------------------------
  *
  * PATTERN: On day 45, the dominant `source` for "post viewed" flips from
@@ -146,7 +146,7 @@ const chance = u.initChance(SEED);
  * quality is awful, dragging down avg watch time.
  *
  * -------------------------------------------------------------------------------------
- * 5. NOTIFICATION RE-ENGAGEMENT (event)
+ * 5. NOTIFICATION RE-ENGAGEMENT (everything)
  * -------------------------------------------------------------------------------------
  *
  * PATTERN: After day 30, 30% of "post viewed" events have source flipped
@@ -190,14 +190,14 @@ const chance = u.initChance(SEED);
  * 7. TOXICITY CHURN (everything)
  * -------------------------------------------------------------------------------------
  *
- * PATTERN: Users with 3+ "report submitted" events lose 60% of activity
+ * PATTERN: Users with 2+ "report submitted" events lose 60% of activity
  * after day 30. No flag — discover via retention or per-user activity drop.
  *
  * HOW TO FIND IT IN MIXPANEL:
  *
  *   Report 1: Retention by Toxicity
  *   - Report type: Retention
- *   - Cohort A: users with >= 3 "report submitted"
+ *   - Cohort A: users with >= 2 "report submitted"
  *   - Cohort B: rest
  *   - Expected: A ~ 40% retention vs B ~ 80%
  *
@@ -276,15 +276,17 @@ const postIds = v.range(1, 1001).map(n => `post_${v.uid(8)}`);
 
 /** @type {Config} */
 const config = {
+	version: 2,
 	token,
 	seed: SEED,
 	datasetStart: "2026-01-01T00:00:00Z",
-	datasetEnd: "2026-04-28T23:59:59Z",
+	datasetEnd: "2026-05-01T23:59:59Z",
 	soup: { dayOfWeekWeights: [1.0, 1.0, 1.0, 1.0, 1.0, 1.2, 1.2] },
 	// numDays: num_days,
 	avgEventsPerUserPerDay: avg_events_per_user_per_day,
 	numUsers: num_users,
-	hasAnonIds: false,
+	hasAnonIds: true,
+	avgDevicePerUser: 2,
 	hasSessionIds: true,
 	format: "json",
 	gzip: true,
@@ -379,6 +381,7 @@ const config = {
 			event: "account created",
 			weight: 1,
 			isFirstEvent: true,
+			isAuthEvent: true,
 			properties: {
 				"signup_method": ["email", "google", "apple", "sso"],
 				"referred_by": ["organic", "friend", "ad", "influencer"],
@@ -576,43 +579,6 @@ const config = {
 		}
 
 
-		// ─── EVENT-LEVEL HOOKS ───────────────────────────────────────────
-
-		if (type === "event") {
-			const datasetStart = dayjs.unix(meta.datasetStart);
-			const ALGORITHM_CHANGE_DAY = datasetStart.add(45, 'days');
-			const REENGAGEMENT_START = datasetStart.add(30, 'days');
-			const EVENT_TIME = dayjs(record.time);
-
-			// Hook #3: ALGORITHM CHANGE - Day 45 flips feed -> explore.
-			// Mutates the existing config-defined `source` prop.
-			if (record.event === "post viewed") {
-				if (EVENT_TIME.isAfter(ALGORITHM_CHANGE_DAY)) {
-					if (chance.bool({ likelihood: 70 })) {
-						record.source = "explore";
-					}
-				} else {
-					if (chance.bool({ likelihood: 70 })) {
-						record.source = "feed";
-					}
-				}
-			}
-
-			// Hook #4: ENGAGEMENT BAIT - 20% of post views get crushed view duration.
-			// No flag — analyst sees bimodal duration distribution + low-tail share.
-			if (record.event === "post viewed") {
-				if (chance.bool({ likelihood: 20 })) {
-					record.view_duration_sec = chance.integer({ min: 1, max: 5 });
-				}
-
-				// Hook #5: NOTIFICATION RE-ENGAGEMENT — after day 30, 30% of views
-				// flip source to "notification". Mutates existing source prop.
-				if (EVENT_TIME.isAfter(REENGAGEMENT_START) && chance.bool({ likelihood: 30 })) {
-					record.source = "notification";
-				}
-			}
-		}
-
 		// ─── EVERYTHING-LEVEL HOOKS ──────────────────────────────────────
 
 		if (type === "everything") {
@@ -782,9 +748,42 @@ const config = {
 				}
 			}
 
+			// Hook #3: ALGORITHM CHANGE — day 45 flips feed → explore on post viewed.
+			// Hook #4: ENGAGEMENT BAIT — 20% of post-viewed events get crushed duration.
+			// Hook #5: NOTIFICATION RE-ENGAGEMENT — after day 30, 30% of views → notification.
+			// All three run AFTER injection passes so they apply to cloned events too.
+			const ALGORITHM_CHANGE_DAY = datasetStart.add(45, 'days');
+			const REENGAGEMENT_START = datasetStart.add(30, 'days');
+			userEvents.forEach(e => {
+				if (e.event === "post viewed") {
+					const eventTime = dayjs(e.time);
+
+					// Hook #3: Algorithm Change
+					if (eventTime.isAfter(ALGORITHM_CHANGE_DAY)) {
+						if (chance.bool({ likelihood: 70 })) {
+							e.source = "explore";
+						}
+					} else {
+						if (chance.bool({ likelihood: 70 })) {
+							e.source = "feed";
+						}
+					}
+
+					// Hook #4: Engagement Bait — 20% crushed view duration
+					if (chance.bool({ likelihood: 20 })) {
+						e.view_duration_sec = chance.integer({ min: 1, max: 5 });
+					}
+
+					// Hook #5: Notification Re-engagement (runs after #3 so can override)
+					if (eventTime.isAfter(REENGAGEMENT_START) && chance.bool({ likelihood: 30 })) {
+						e.source = "notification";
+					}
+				}
+			});
+
 			// Hook #7: TOXICITY CHURN — drop 60% of activity after day 30 for high reporters.
-			// Discovery: cohort users with >=3 report-submitted events, observe retention drop.
-			if (reportSubmittedCount >= 3) {
+			// Discovery: cohort users with >=2 report-submitted events, observe retention drop.
+			if (reportSubmittedCount >= 2) {
 				const churnCutoff = datasetStart.add(30, 'days');
 				for (let i = userEvents.length - 1; i >= 0; i--) {
 					const evt = userEvents[i];

package/dungeons/vertical/travel.js

@@ -1,7 +1,7 @@
 // ── TWEAK THESE ──
 const SEED = "dm4-travel";
-const num_days = 100;
-const num_users = 5_000;
+const num_days = 120;
+const num_users = 10_000;
 const avg_events_per_user_per_day = 1.2;
 let token = "your-mixpanel-token";
 
@@ -29,7 +29,7 @@ const destinationCities = ["New York", "London", "Paris", "Tokyo", "Barcelona",
  * StayQuest — a hotel booking platform for business and leisure travelers.
  * Users search destinations, compare hotels, book rooms, and leave reviews.
  *
- * - 5,000 users over 100 days, ~600K events
+ * - 5,000 users over 120 days, ~600K events
  * - Segments: business travelers (weekday), leisure families, luxury, budget
  * - Core loop: search → view hotel → compare → book → stay → review
  * - Revenue: commission per booking + premium loyalty membership
@@ -210,14 +210,16 @@ const destinationCities = ["New York", "London", "Paris", "Tokyo", "Barcelona",
 
 /** @type {Config} */
 const config = {
+	version: 2,
 	token,
 	seed: SEED,
 	datasetStart: "2026-01-01T00:00:00Z",
-	datasetEnd: "2026-04-28T23:59:59Z",
+	datasetEnd: "2026-05-01T23:59:59Z",
 	// numDays: num_days,
 	avgEventsPerUserPerDay: avg_events_per_user_per_day,
 	numUsers: num_users,
-	hasAnonIds: false,
+	hasAnonIds: true,
+	avgDevicePerUser: 2,
 	hasSessionIds: true,
 	format: "json",
 	gzip: true,
@@ -250,6 +252,7 @@ const config = {
 			event: "account created",
 			weight: 1,
 			isFirstEvent: true,
+			isAuthEvent: true,
 			properties: {
 				signup_source: ["organic", "google", "instagram", "tripadvisor", "referral", "email_campaign"],
 			},

package/index.js

@@ -18,14 +18,14 @@ import { detectInputType, loadFromFile, loadFromText, parseJSONDungeon, validate
 
 // Orchestrators
 import { userLoop } from './lib/orchestrators/user-loop.js';
-import { sendToMixpanel } from './lib/orchestrators/mixpanel-sender.js';
+import { sendToMixpanel, collectWrittenFiles } from './lib/orchestrators/mixpanel-sender.js';
 // Generators
 import { makeAdSpend } from './lib/generators/adspend.js';
 import { makeMirror } from './lib/generators/mirror.js';
 import { makeGroupProfile, makeProfile } from './lib/generators/profiles.js';
 
 // Utilities
-import { initChance, setDatasetNow } from './lib/utils/utils.js';
+import { initChance, setDatasetNow, deleteFile } from './lib/utils/utils.js';
 
 // External dependencies
 import dayjs from "dayjs";
@@ -122,6 +122,7 @@ async function runDungeon(config) {
 
 	if (config.verbose) logger.info({ seed: config.seed }, 'Configuring dungeon');
 	let validatedConfig;
+	let storage;
 	try {
 		// Initialize seeded RNG BEFORE validation — config-validator captures a
 		// chance reference for default userProps (spiritAnimal). If we init after,
@@ -148,7 +149,7 @@ async function runDungeon(config) {
 
 		// Step 3: Initialize storage containers
 		const storageManager = new StorageManager(context);
-		const storage = await storageManager.initializeContainers();
+		storage = await storageManager.initializeContainers();
 		updateContextWithStorage(context, storage);
 
 		// ! DATA GENERATION STARTS HERE
@@ -215,7 +216,7 @@ async function runDungeon(config) {
 		return {
 			...extractedData,
 			importResults,
-			files: await extractFileInfo(storage, validatedConfig),
+			files: extractFileInfo(storage),
 			time: { start, end, delta, human },
 			operations: context.getOperations(),
 			eventCount: context.getStoredEventCount(),
@@ -225,6 +226,15 @@ async function runDungeon(config) {
 	} catch (error) {
 		logger.error({ err: error }, `Error: ${error.message}`);
 		throw error;
+	} finally {
+		if (validatedConfig?.cleanup && storage) {
+			const allFiles = collectWrittenFiles(storage);
+			if (allFiles.length > 0) {
+				if (validatedConfig.verbose) console.log(`\nCleaning up ${allFiles.length} written files...`);
+				await Promise.allSettled(allFiles.map(f => deleteFile(f)));
+				if (validatedConfig.verbose) console.log(`Cleanup complete`);
+			}
+		}
 	}
 }
 
@@ -480,74 +490,10 @@ async function flushStorageToDisk(storage, config) {
 /**
  * Extract file information from storage containers
  * @param {import('./types').Storage} storage - Storage object
- * @param {import('./types').Dungeon} config - Configuration object
- * @returns {Promise<string[]>} Array of file paths
+ * @returns {string[]} Array of file paths
  */
-async function extractFileInfo(storage, config) {
-	const files = [];
-
-	// Try to get paths from containers first
-	Object.values(storage).forEach(container => {
-		if (Array.isArray(container)) {
-			container.forEach(subContainer => {
-				if (subContainer?.getWritePath) {
-					files.push(subContainer.getWritePath());
-				}
-			});
-		} else if (container?.getWritePath) {
-			files.push(container.getWritePath());
-		}
-	});
-
-	// If no files found from containers and writeToDisk is enabled, scan the data directory
-	if (files.length === 0 && config.writeToDisk) {
-		try {
-			const fs = await import('fs');
-			const path = await import('path');
-			
-			let dataDir = path.resolve("./data");
-			if (!fs.existsSync(dataDir)) {
-				dataDir = path.resolve("./");
-			}
-			
-			if (fs.existsSync(dataDir)) {
-				const allFiles = fs.readdirSync(dataDir);
-				const simulationName = config.name;
-				
-				// Filter files that match our patterns and were likely created by this run
-				const relevantFiles = allFiles.filter(file => {
-					// Skip system files
-					if (file.startsWith('.')) return false;
-					
-					// If we have a simulation name, only include files with that prefix
-					if (simulationName && !file.startsWith(simulationName)) {
-						return false;
-					}
-					
-					// Check for common patterns
-					const hasEventPattern = file.includes('-EVENTS.');
-					const hasUserPattern = file.includes('-USERS.');
-					const hasScdPattern = file.includes('-SCD.');
-					const hasGroupPattern = file.includes('-GROUPS.');
-					const hasLookupPattern = file.includes('-LOOKUP.');
-					const hasAdspendPattern = file.includes('-ADSPEND.');
-					const hasMirrorPattern = file.includes('-MIRROR.');
-					
-					return hasEventPattern || hasUserPattern || hasScdPattern || 
-						   hasGroupPattern || hasLookupPattern || hasAdspendPattern || hasMirrorPattern;
-				});
-				
-				// Convert to full paths
-				relevantFiles.forEach(file => {
-					files.push(path.join(dataDir, file));
-				});
-			}
-		} catch (error) {
-			// If scanning fails, just return empty array
-		}
-	}
-
-	return files;
+function extractFileInfo(storage) {
+	return collectWrittenFiles(storage);
 }
 
 /**