@codeyam/codeyam-cli 0.1.0-staging.bd4126b → 0.1.0-staging.c21fa76

package/analyzer-template/packages/ai/src/lib/dataStructure/ScopeDataStructure.ts

@@ -398,6 +398,7 @@ const SILENTLY_IGNORED_EQUIVALENCY_REASONS = new Set([
   'transformed non-object function equivalency - Array.from() equivalency',
   'Spread operator equivalency key update: Explicit array deconstruction equivalency value',
   // 'transformed non-object function equivalency - Explicit array deconstruction equivalency value',
+  'useState setter call equivalency (direct)',
 ]);
 
 export class ScopeDataStructure {
@@ -1048,9 +1049,38 @@ export class ScopeDataStructure {
     equivalencyValueChain?: EquivalencyValueChainItem[],
     traceId?: number,
   ) {
+    // Fast early exits that don't count toward the limit.
+    // These are "no-op" calls that would be caught further below anyway,
+    // but checking them first prevents counter inflation from redundant calls
+    // generated by managers and followEquivalencies during scope processing.
+
+    // Skip silently-ignored reasons before counting
+    if (SILENTLY_IGNORED_EQUIVALENCY_REASONS.has(equivalencyReason)) {
+      return;
+    }
+
+    // Skip exact duplicates before counting (Phase 1 only — Phase 2 needs
+    // to process duplicates for database chain tracking)
+    if (this.onlyEquivalencies) {
+      const existing = scopeNode.equivalencies[path];
+      if (
+        existing?.some(
+          (v) =>
+            v.schemaPath === equivalentPath &&
+            v.scopeNodeName === equivalentScopeName,
+        )
+      ) {
+        return;
+      }
+    }
+
     addEquivalencyCallCount++;
 
-    if (addEquivalencyCallCount > 50000) {
+    const limit =
+      typeof (globalThis as any).__CODEYAM_EQUIVALENCY_LIMIT === 'number'
+        ? (globalThis as any).__CODEYAM_EQUIVALENCY_LIMIT
+        : 50000;
+    if (addEquivalencyCallCount > limit) {
       console.error('INFINITE LOOP DETECTED in addEquivalency', {
         callCount: addEquivalencyCallCount,
         path,
@@ -1065,14 +1095,11 @@ export class ScopeDataStructure {
     }
     // Filter equivalency reasons - use pre-computed Sets for O(1) lookup
     if (!ALLOWED_EQUIVALENCY_REASONS.has(equivalencyReason)) {
-      if (SILENTLY_IGNORED_EQUIVALENCY_REASONS.has(equivalencyReason)) {
-        return;
-      } else {
-        // Log and skip - if an equivalency reason isn't in ALLOWED or SILENTLY_IGNORED,
-        // it shouldn't be stored (was previously missing the return)
-        console.info('Not tracked equivalency reason', { equivalencyReason });
-        return;
-      }
+      // Silently-ignored reasons already handled above
+      // Log and skip - if an equivalency reason isn't in ALLOWED or SILENTLY_IGNORED,
+      // it shouldn't be stored (was previously missing the return)
+      console.info('Not tracked equivalency reason', { equivalencyReason });
+      return;
     }
 
     // Skip self-referential equivalencies that would create infinite loops.
@@ -1108,6 +1135,33 @@ export class ScopeDataStructure {
       return;
     }
 
+    // Case 3: Circular reference through scope-suffixed names (____cyScope pattern)
+    // When a named arrow function is defined inside a scope (e.g., useEffect callback):
+    //   const identifyUser = async () => { ... };
+    //   identifyUser();
+    // This creates a variable "identifyUser" and a scope "identifyUser____cyScope9F".
+    // Mutual equivalencies between these cause infinite loops in Phase 2 because
+    // processing one triggers addToSchema → followEquivalencies → addEquivalency
+    // on the reverse, which repeats indefinitely.
+    // Only block when the REVERSE direction already exists (creating a cycle).
+    // The initial one-directional equivalency is necessary for scope resolution.
+    if (
+      path &&
+      equivalentPath &&
+      (equivalentPath.startsWith(path + '____') ||
+        path.startsWith(equivalentPath + '____'))
+    ) {
+      // Check if the reverse equivalency already exists
+      const reverseEquivalencies =
+        scopeNode.equivalencies[equivalentPath] || [];
+      const reverseExists = reverseEquivalencies.some(
+        (v) => v.schemaPath === path,
+      );
+      if (reverseExists) {
+        return;
+      }
+    }
+
     if (!equivalentScopeName) {
       console.error(
         'CodeYam Error: Missing equivalent scope name - FULL CONTEXT:',

package/analyzer-template/packages/ai/src/lib/dataStructure/equivalencyManagers/ParentScopeManager.ts

@@ -97,6 +97,16 @@ export default class ParentScopeManager implements EquivalencyManager {
       }
 
       if (instantiatedInParent || returnValueFromChildScope) {
+        // Skip if we've already propagated this variable to the parent scope.
+        // Without this guard, nested scopes trigger redundant addEquivalency calls
+        // for the same variable on every addToSchema pass (e.g., 'a' and 'b' in sort
+        // callbacks get propagated 30+ times per scope instead of once).
+        this.addedToParent[scopeNode.name] ||= new Set<string>();
+        if (this.addedToParent[scopeNode.name].has(rootPath)) {
+          return false;
+        }
+        this.addedToParent[scopeNode.name].add(rootPath);
+
         if (traceId && debugLevel > 0) {
           console.info(
             'Debug Propagation: parentScope executed',
@@ -114,9 +124,6 @@ export default class ParentScopeManager implements EquivalencyManager {
           );
         }
 
-        this.addedToParent[scopeNode.name] ||= new Set<string>();
-        this.addedToParent[scopeNode.name].add(rootPath);
-
         scopeDataStructure.addEquivalency(
           rootPath,
           rootPath,

package/analyzer-template/packages/ai/src/lib/dataStructure/helpers/cleanKnownObjectFunctions.ts

@@ -6,7 +6,7 @@ import { ScopeNode, ScopeDataStructure } from '../ScopeDataStructure';
 
 type PrimitiveKind = 'array' | 'string' | 'number' | 'boolean' | 'date';
 
-const IGNORE_CLASSES_AND_OBJECTS = [
+const IGNORE_CLASSES_AND_OBJECTS_SET = new Set([
   'Math',
   'JSON',
   'Reflect',
@@ -22,7 +22,7 @@ const IGNORE_CLASSES_AND_OBJECTS = [
   'useContext',
   'useReducer',
   'useLayoutEffect',
-];
+]);
 
 /** cache the tokenised representation of every path we touch */
 const partsCache = new Map<string, string[]>();
@@ -147,16 +147,26 @@ function scrub(
   rootTable: Map<string, PrimitiveKind>,
   onRemove: (key: string) => void,
 ): void {
-  const sortedKeys = Object.keys(mapping).sort((a, b) => a.localeCompare(b));
+  // Early exit: if no primitive types exist in the schema, there are no
+  // built-in methods to clean. Saves sorting + walking 10K+ keys.
+  if (rootTable.size === 0) return;
+
+  // Use direct comparison instead of localeCompare (3-5x faster for large arrays)
+  const sortedKeys = Object.keys(mapping).sort((a, b) =>
+    a < b ? -1 : a > b ? 1 : 0,
+  );
 
   for (const keyPath of sortedKeys) {
     const keyParts = getParts(keyPath);
     const prefixParts = buildPrefixParts(keyParts);
 
+    const firstPart = keyParts[0];
+    const firstPartBase = firstPart.includes('(')
+      ? firstPart.split('(')[0]
+      : firstPart;
     if (
-      IGNORE_CLASSES_AND_OBJECTS.find(
-        (m) => keyParts[0] === m || keyParts[0].startsWith(`${m}(`),
-      )
+      IGNORE_CLASSES_AND_OBJECTS_SET.has(firstPart) ||
+      IGNORE_CLASSES_AND_OBJECTS_SET.has(firstPartBase)
     ) {
       delete mapping[keyPath];
       continue;

package/analyzer-template/packages/ai/src/lib/dataStructure/helpers/coercePrimitivesToArraysBySchema.ts

@@ -0,0 +1,62 @@
+/**
+ * Coerces primitive values to empty arrays when the schema expects an array.
+ *
+ * The LLM sometimes generates a primitive (boolean, number, string) where the
+ * schema expects an array. For example:
+ *   Schema: { webapps: [{ name: "string" }] }
+ *   LLM output: { webapps: true }
+ *
+ * This causes runtime errors like `TypeError: config.webapps?.forEach is not a function`
+ * because the code expects an array but gets a boolean.
+ *
+ * Coercion rule:
+ * - primitive → array: replace with []
+ *
+ * Null and undefined values are preserved (they may be intentionally nullable).
+ *
+ * @param data The mock data object to coerce (mutated in place)
+ * @param schema The schema describing expected types
+ */
+export default function coercePrimitivesToArraysBySchema<T>(
+  data: T,
+  schema: unknown,
+): T {
+  if (data === null || data === undefined) {
+    return data;
+  }
+
+  if (typeof data !== 'object') {
+    return data;
+  }
+
+  if (Array.isArray(data)) {
+    const itemSchema = Array.isArray(schema) ? schema[0] : undefined;
+    for (let i = 0; i < data.length; i++) {
+      data[i] = coercePrimitivesToArraysBySchema(data[i], itemSchema);
+    }
+    return data;
+  }
+
+  const dataObj = data as Record<string, unknown>;
+  const schemaObj = schema as Record<string, unknown> | undefined;
+
+  for (const key of Object.keys(dataObj)) {
+    const value = dataObj[key];
+    const fieldSchema = schemaObj?.[key];
+
+    if (value === null || value === undefined) {
+      continue;
+    }
+
+    if (typeof value !== 'object' && Array.isArray(fieldSchema)) {
+      // Schema expects an array but the value is a primitive.
+      // Replace with an empty array.
+      dataObj[key] = [];
+    } else if (typeof value === 'object' && value !== null) {
+      // Both value and schema are objects — recurse
+      coercePrimitivesToArraysBySchema(value, fieldSchema);
+    }
+  }
+
+  return data;
+}

package/analyzer-template/packages/ai/src/lib/generateEntityScenarioData.ts

@@ -19,6 +19,8 @@ import convertNullToUndefinedBySchema from './dataStructure/helpers/convertNullT
 import convertTypeAnnotationsToValues from './dataStructure/helpers/convertTypeAnnotationsToValues';
 import fixNullIdsBySchema from './dataStructure/helpers/fixNullIdsBySchema';
 import coerceObjectsToPrimitivesBySchema from './dataStructure/helpers/coerceObjectsToPrimitivesBySchema';
+import coercePrimitivesToArraysBySchema from './dataStructure/helpers/coercePrimitivesToArraysBySchema';
+import stripNullableMarkers from './dataStructure/helpers/stripNullableMarkers';
 import { JsonTypeDefinition } from '~codeyam/types';
 import { deepMerge } from '~codeyam/generate';
 import {
@@ -172,6 +174,46 @@ function findKeyPath(
   return null;
 }
 
+/**
+ * Strip primitive values from `source` when the `destination` already has an
+ * array or object at the same key. Returns a cleaned copy of source (or null
+ * if nothing remains after stripping).
+ *
+ * This prevents misplaced LLM data (e.g., `{ webapps: true }`) from
+ * overwriting correct complex values (e.g., `{ webapps: [...] }`) during
+ * deep merge in relocateMisplacedNestedKeys.
+ */
+function stripPrimitivesOverwritingComplexValues(
+  destination: Record<string, unknown>,
+  source: unknown,
+): Record<string, unknown> | null {
+  if (typeof source !== 'object' || source === null || Array.isArray(source)) {
+    return null;
+  }
+
+  const result: Record<string, unknown> = {};
+  let hasKeys = false;
+
+  for (const key of Object.keys(source as Record<string, unknown>)) {
+    const srcVal = (source as Record<string, unknown>)[key];
+    const dstVal = destination[key];
+
+    // If destination has an array or object but source has a primitive, skip it
+    if (
+      typeof dstVal === 'object' &&
+      dstVal !== null &&
+      typeof srcVal !== 'object'
+    ) {
+      continue;
+    }
+
+    result[key] = srcVal;
+    hasKeys = true;
+  }
+
+  return hasKeys ? result : null;
+}
+
 /**
  * Relocate misplaced nested keys in mockData to their correct position
  * based on the dataForMocks structure.
@@ -233,12 +275,22 @@ function relocateMisplacedNestedKeys(
     }
 
     // Deep merge the value into the correct location
-    // Use deep merge to preserve existing data at that location
+    // Use deep merge to preserve existing data at that location.
+    // Before merging, strip primitives from the misplaced value that would
+    // overwrite arrays/objects at the destination. The misplaced data is lower
+    // quality — the LLM put it in the wrong place — so primitives like `true`
+    // should not overwrite correct complex values like arrays.
     if (current[key] !== undefined && typeof current[key] === 'object') {
-      current[key] = deepMerge(
+      const safeValue = stripPrimitivesOverwritingComplexValues(
         current[key] as Record<string, unknown>,
         value as Record<string, unknown>,
       );
+      if (safeValue !== null) {
+        current[key] = deepMerge(
+          current[key] as Record<string, unknown>,
+          safeValue,
+        );
+      }
     } else {
       current[key] = value;
     }
@@ -1063,6 +1115,18 @@ export async function generateDataForScenario({
     );
   }
 
+  // Strip _nullable markers from LLM-generated mock data.
+  // The _nullable marker is an internal CodeYam concept used in type definitions
+  // to indicate that a field can be null/undefined. The LLM sometimes includes
+  // these markers in its generated data, which causes runtime errors when code
+  // iterates over object keys (e.g., Object.keys(importedBy) picks up "_nullable",
+  // then Object.keys(importedBy["_nullable"]) calls Object.keys(null) and throws).
+  if (fullScenarioData.data.mockData) {
+    stripNullableMarkers(
+      fullScenarioData.data.mockData as Record<string, unknown>,
+    );
+  }
+
   // Relocate misplaced nested keys to their correct position.
   // The LLM sometimes places nested keys at root level instead of inside their
   // parent object (e.g., 'fastener' at root instead of inside 'trpc').
@@ -1098,6 +1162,18 @@ export async function generateDataForScenario({
     );
   }
 
+  // Coerce primitives to empty arrays when the schema expects an array.
+  // The LLM sometimes generates a primitive (e.g., `webapps: true`) where the
+  // schema expects an array (e.g., `webapps: [{ name: "string" }]`).
+  // This causes runtime errors like "TypeError: config.webapps?.forEach is not a function".
+  // Must run AFTER coerceObjectsToPrimitivesBySchema and BEFORE convertCommaSeparatedStringsToArrays.
+  if (structure.dataForMocks && fullScenarioData.data.mockData) {
+    coercePrimitivesToArraysBySchema(
+      fullScenarioData.data.mockData,
+      structure.dataForMocks,
+    );
+  }
+
   // Convert comma-separated strings to arrays when appropriate.
   // The LLM sometimes generates strings like "color,size" instead of arrays
   // like ["color", "size"] when the schema type is incorrectly inferred as

package/analyzer-template/packages/ai/src/lib/generateExecutionFlows.ts

@@ -129,19 +129,6 @@ export default function generateExecutionFlows({
     fullToShortPathMap[fullPath] = shortPath;
   }
 
-  console.log(
-    `[generateExecutionFlows] ${entity.name}: attributesMap has ${Object.keys(resolvedAttributesMap).length} keys, fullToShortPathMap has ${Object.keys(fullToShortPathMap).length} entries`,
-  );
-  console.log(
-    `[generateExecutionFlows] ${entity.name}: attributesMap keys: [${Object.keys(resolvedAttributesMap).join(', ')}]`,
-  );
-  console.log(
-    `[generateExecutionFlows] ${entity.name}: fullToShortPathMap: ${JSON.stringify(fullToShortPathMap)}`,
-  );
-  console.log(
-    `[generateExecutionFlows] ${entity.name}: equivalentSignatureVariables: ${JSON.stringify(equivalentSignatureVariables)}`,
-  );
-
   // Get pre-computed conditional data from metadata
   const conditionalUsages: Record<string, ConditionalUsage[]> =
     entity.metadata?.isolatedDataStructure?.conditionalUsages ?? {};
@@ -186,26 +173,6 @@ export default function generateExecutionFlows({
     }
   }
 
-  console.log(
-    `[generateExecutionFlows] ${entity.name}: conditionalUsages keys: [${Object.keys(conditionalUsages).join(', ')}]`,
-  );
-  console.log(
-    `[generateExecutionFlows] ${entity.name}: conditionalUsages sourceDataPaths: ${JSON.stringify(
-      Object.fromEntries(
-        Object.entries(conditionalUsages).map(([k, v]) => [
-          k,
-          (v as any[]).map((u: any) => u.sourceDataPath ?? '(none)'),
-        ]),
-      ),
-    )}`,
-  );
-  console.log(
-    `[generateExecutionFlows] ${entity.name}: compoundConditionals count: ${compoundConditionals.length}`,
-  );
-  console.log(
-    `[generateExecutionFlows] ${entity.name}: sourceEquivalencies keys count: ${Object.keys(mergedDataStructure.sourceEquivalencies ?? {}).length}`,
-  );
-
   const conditionalsBasedFlows = generateExecutionFlowsFromConditionals({
     conditionalUsages,
     compoundConditionals,

package/analyzer-template/packages/analyze/index.ts

@@ -36,7 +36,10 @@ export { getAllEntityNodes } from './src/lib/asts/sourceFiles/getAllEntityNodes'
 export { default as fileAnalyzerFromCode } from './src/lib/files/fileAnalyzerFromCode';
 
 export { default as isolateDataStructure } from './src/lib/files/scenarios/isolateDataStructure';
-export { default as mergeInDependentDataStructure } from './src/lib/files/scenarios/mergeInDependentDataStructure';
+export {
+  default as mergeInDependentDataStructure,
+  DataStructureTimeoutError,
+} from './src/lib/files/scenarios/mergeInDependentDataStructure';
 export { default as mergeValidatedDataStructures } from './src/lib/files/scenarios/mergeValidatedDataStructures';
 
 export { discoverDirectDependencies } from './src/lib/files/analyze/dependencyResolver';

package/analyzer-template/packages/analyze/src/lib/ProjectAnalyzer.ts

@@ -46,6 +46,12 @@ export class ProjectAnalyzer {
       lib.asts.createProgramFromTsConfigPath(configPath, excludePatterns),
     );
 
+    if (this.programs.length === 0) {
+      throw new Error(
+        'No TypeScript programs could be created — no tsconfig.json found',
+      );
+    }
+
     // Primary program is the first one (usually root) for backward compatibility
     this.program = this.programs[0];
     this.typeChecker = this.program.getTypeChecker();
@@ -334,17 +340,23 @@ export class ProjectAnalyzer {
    * Refresh the TypeScript program to pick up file changes
    */
   refreshProgram(): void {
-    // Memory logging: Capture state before refresh
-    const memBefore = process.memoryUsage();
-
     // Apply the same unapprovedPaths filtering when refreshing
     const excludePatterns =
       this.project.metadata?.unapprovedPaths?.filter(Boolean) || [];
 
-    // Refresh all programs
-    this.programs = this.tsConfigPaths.map((configPath) =>
-      lib.asts.createProgramFromTsConfigPath(configPath, excludePatterns),
-    );
+    // Release old programs and caches BEFORE creating new ones to reduce peak memory.
+    // Without this, both old (~1.7GB) and new (~1.7GB) programs coexist during creation,
+    // causing OOM on monorepos with multiple tsconfigs.
+    this.programs = [];
+    this.sourceFileByPath.clear();
+    this.typeCheckerByPath.clear();
+
+    // Create programs one at a time so each old program can be GC'd before the next
+    for (const configPath of this.tsConfigPaths) {
+      this.programs.push(
+        lib.asts.createProgramFromTsConfigPath(configPath, excludePatterns),
+      );
+    }
 
     // Update primary program for backward compatibility
     this.program = this.programs[0];

package/analyzer-template/packages/analyze/src/lib/asts/index.ts

@@ -95,8 +95,13 @@ export function createProgramFromTsConfigPath(
     throw new Error('Could not parse tsconfig.json');
   }
 
-  // Filter file names based on exclude patterns to reduce memory usage
-  let fileNames = parsedCommandLine.fileNames;
+  // Always filter out node_modules from root files to prevent OOM on large projects.
+  // This only removes them as root compilation targets — TypeScript's module resolution
+  // still follows imports into node_modules for type info when processing source files.
+  let fileNames = parsedCommandLine.fileNames.filter(
+    (f) => !f.includes('/node_modules/'),
+  );
+
   if (excludePatterns && excludePatterns.length > 0) {
     const regexPatterns = excludePatterns
       .map((pattern) => {

package/analyzer-template/packages/analyze/src/lib/files/analyze/analyzeEntities/prepareDataStructures.ts

@@ -289,18 +289,44 @@ export default async function prepareEntityDataStructures(
     skipWorkerPool();
   }
 
+  awsLogDebugLevel(
+    1,
+    `Preparing data structures for ${entities.length} entities (${entitiesNeedingGeneration} need generation)...`,
+  );
+  const startTime = Date.now();
+  let completedCount = 0;
+
+  // Wrap processEntity to track progress
+  const processEntityWithProgress = async (entity: Entity) => {
+    await processEntity(entity);
+    completedCount++;
+    // Emit progress every 10 entities
+    if (completedCount % 10 === 0) {
+      awsLogDebugLevel(
+        1,
+        `Preparing data structures... (${completedCount}/${entities.length})`,
+      );
+    }
+  };
+
   const sequential =
     DEBUG_SEQUENTIAL_EXECUTION || options.sequentialDataStructurePrep;
 
   if (sequential) {
     // Sequential execution for debugging - process one entity at a time
     for (const entity of entities) {
-      await processEntity(entity);
+      await processEntityWithProgress(entity);
     }
   } else {
     // Parallel execution for speed
-    await Promise.all(entities.map(processEntity));
+    await Promise.all(entities.map(processEntityWithProgress));
   }
+
+  const elapsed = Date.now() - startTime;
+  awsLogDebugLevel(
+    1,
+    `Data structure preparation complete: ${entities.length} entities in ${elapsed}ms`,
+  );
 }
 
 /**

package/analyzer-template/packages/analyze/src/lib/files/analyze/analyzeEntities.ts

@@ -8,13 +8,11 @@ import {
 import { transformationTracer } from '../scenarios/TransformationTracer';
 import { getFileByPathSafe } from '../../utils/getFileByPath';
 import trackEntityCircularDependencies from './trackEntityCircularDependencies';
-import validateDependencyAnalyses from './validateDependencyAnalyses';
 import extractClassMethods from './analyzeEntities/extractClassMethods';
 import prepareDataStructures from './analyzeEntities/prepareDataStructures';
 import { strategies, asyncComplex } from './analyzeEntities/strategies';
 import {
   updateCommitMetadata,
-  updateFreshAnalysisMetadata,
   updateFreshAnalysisStatus,
   upsertAnalyses,
   upsertEntities,
@@ -263,40 +261,11 @@ export default async function analyzeEntities({
     }
   }
 
-  awsLog(`CodeYam: analyzeEntities: Running patch-up dependency validation`);
-  for (const analysis of analyses) {
-    if (!analysis.entity) {
-      throw new Error(
-        `CodeYam: analyzeEntities: Internal error: Analysis entity is missing for ${analysis.filePath}:${analysis.entityName}.`,
-      );
-    }
-
-    // Validate dependencies, which modifies analysis.metadata.dependentAnalyses
-    validateDependencyAnalyses({ analysis, entity: analysis.entity, context });
-
-    // Carefully update only the metadata field
-    if (analysis.id) {
-      const updatedMetadata = await updateFreshAnalysisMetadata(
-        analysis.id,
-        (metadata) => {
-          metadata.dependentAnalyses =
-            analysis.metadata?.dependentAnalyses || [];
-        },
-      );
-
-      if (updatedMetadata) {
-        analysis.metadata = updatedMetadata;
-      } else {
-        console.warn(
-          `CodeYam: analyzeEntities: Failed to update metadata for analysis ${analysis.id}`,
-        );
-      }
-    } else {
-      console.warn(
-        `CodeYam: analyzeEntities: Analysis ${analysis.filePath}:${analysis.entityName} has no ID, skipping metadata update`,
-      );
-    }
-  }
+  // NOTE: patch-up dependency validation loop was removed here.
+  // analysis.metadata.dependentAnalyses is already populated by validateDependencyAnalyses
+  // inside analyzeEntity() (Call #1) and persisted via upsertAnalysesWithScenarios at the
+  // end of analyzeEntity, then again by upsertAnalyses in analyzeOneEntityKey. Re-validating
+  // every analysis from scratch was redundant work — especially costly for large projects.
 
   const newlyReady = await updateReadyAnalyses();
   if (newlyReady.length > 0) {
@@ -310,7 +279,15 @@ export default async function analyzeEntities({
   );
 
   // Flush transformation trace if tracing is enabled
-  transformationTracer.flush();
+  // Wrapped in try/catch because trace data can exceed V8's max string length
+  // for large projects, and we must not crash the analysis process over tracing.
+  try {
+    transformationTracer.flush();
+  } catch (e) {
+    console.warn(
+      `CodeYam Warning: Failed to flush transformation trace: ${e.message}`,
+    );
+  }
 
   return analyses;
 

package/analyzer-template/packages/analyze/src/lib/files/analyze/dependencyResolver.ts

@@ -100,12 +100,6 @@ export function getCandidatesReadyForCapture(
       continue;
     }
 
-    // Skip notExported entities - they can't be imported into the simulation
-    // environment and will always fail capture. Don't mark them as ready.
-    if (candidate.entity?.metadata?.notExported) {
-      continue;
-    }
-
     // Get all transitive dependencies
     const candidateKey = `${candidate.filePath}:${candidate.entityName}`;
     const { missingDependencies } = getTransitiveDependencies(

package/analyzer-template/packages/analyze/src/lib/files/analyze/findOrCreateEntity.ts

@@ -94,12 +94,15 @@ export default async function findOrCreateEntity({
   try {
     code = fileAnalyzer.getEntityCode(localEntityName);
   } catch (e) {
-    console.log('CodeYam Error: could not get entity code', {
-      filePath: file.path,
-      localEntityName,
-      entityName,
-      error: e,
-    });
+    console.log(
+      `CodeYam Error: Could not extract code for "${localEntityName}" from ${file.path} — export not found in file, may have been moved or renamed during refactoring`,
+      {
+        filePath: file.path,
+        localEntityName,
+        entityName,
+        error: String(e),
+      },
+    );
     return null;
   }
   const sha = generateSha(file.path, entityName, code);
@@ -108,6 +111,7 @@ export default async function findOrCreateEntity({
     entity = await loadEntity({
       projectId: project.id,
       sha,
+      silent: true, // Entity not existing yet is the normal path — we create it below
     });
   }
 
@@ -138,6 +142,18 @@ export default async function findOrCreateEntity({
           break;
         }
       }
+
+      // Mark previous versions as superseded
+      const entitiesToSupersede = previousEntities.filter(
+        (e) => !e.metadata?.isSuperseded,
+      );
+      if (entitiesToSupersede.length > 0) {
+        for (const prev of entitiesToSupersede) {
+          prev.metadata = prev.metadata || {};
+          prev.metadata.isSuperseded = true;
+        }
+        await upsertEntities(entitiesToSupersede);
+      }
     }
 
     entity = {