Merge pull request #208 from PredicateSystems/self_improving2

agent: extensible profiles, data-driven pruning, and task learning

Author: rcholic

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@predicatesystems/runtime",
-  "version": "1.3.4",
+  "version": "1.4.0",
   "description": "TypeScript SDK for Sentience AI Agent Browser Automation",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",

src/agents/browser-agent.ts

@@ -207,6 +207,48 @@ class TokenAccountingProvider extends LLMProvider {
   }
 }
 
+// Re-export planner-executor profile/learning types for extension consumption
+export type {
+  DataDrivenPruningPolicy,
+  BrowserAgentProfile,
+  ResolvedAgentProfile,
+  LearnedTargetFingerprint,
+  DomainProfile,
+} from './planner-executor/profile-types';
+export { EMPTY_RESOLVED_PROFILE } from './planner-executor/profile-types';
+export {
+  DataDrivenPruningPolicySchema,
+  BrowserAgentProfileSchema,
+  BrowserAgentProfileArraySchema,
+  LearnedTargetFingerprintSchema,
+  DomainProfileSchema,
+} from './planner-executor/profile-schema';
+export { ProfileRegistry } from './planner-executor/profile-registry';
+export { pruneWithPolicy } from './planner-executor/data-driven-pruner';
+export {
+  computeTaskHash,
+  extractDomain,
+  createFingerprint,
+  mergeFingerprint,
+  recordFingerprintFailure,
+} from './planner-executor/fingerprint-normalizer';
+export type { LearningStore } from './planner-executor/learning-store';
+export { InMemoryLearningStore } from './planner-executor/learning-store';
+export {
+  extractFingerprintFromOutcome,
+  applyFingerprintFailure,
+  applyFingerprintSuccess,
+  isFingerprintStale,
+  isFingerprintExpired,
+  fingerprintToHint,
+  computeTaskHash as computeAsyncTaskHash,
+  isSensitiveUrl as isLearningSensitiveUrl,
+} from './planner-executor/learning-extractor';
+export type {
+  LearningExtractionOptions,
+  LearningExtractionResult,
+} from './planner-executor/learning-extractor';
+
 export type StepOutcome = { stepGoal: string; ok: boolean };
 
 export class PredicateBrowserAgent {

src/agents/planner-executor/category-pruner.ts

@@ -6,6 +6,7 @@ import {
   type PrunedSnapshotContext,
 } from './pruning-types';
 import { TaskCategory } from './task-category';
+import { pruneWithPolicy } from './data-driven-pruner';
 
 function textOf(element: SnapshotElement): string {
   return String(element.text || element.name || '').toLowerCase();
@@ -270,6 +271,39 @@ export function pruneSnapshotForTask(
   options: PruneSnapshotOptions
 ): PrunedSnapshotContext {
   const relaxationLevel = Math.max(0, options.relaxationLevel || 0);
+
+  // Data-driven path: use profile policy if provided
+  if (options.profilePolicy) {
+    const { elements, maxNodes } = pruneWithPolicy(
+      snapshot,
+      options.profilePolicy,
+      options.goal,
+      relaxationLevel,
+      options.category,
+      options.learnedFingerprints
+    );
+    const actionableElementCount = selectContextElements(elements, elements.length || 1).length;
+
+    return {
+      category: options.category,
+      snapshot,
+      elements,
+      promptBlock: formatPrunedContext({
+        category: options.category,
+        elements,
+        relaxationLevel,
+        rawElementCount: snapshot.elements.length,
+        prunedElementCount: elements.length,
+        actionableElementCount,
+      }),
+      relaxationLevel,
+      rawElementCount: snapshot.elements.length,
+      prunedElementCount: elements.length,
+      actionableElementCount,
+    };
+  }
+
+  // Built-in category path
   const policy = getPolicy(options.category, relaxationLevel);
   const filtered = (snapshot.elements || []).filter(element => {
     if (policy.block(element)) {

src/agents/planner-executor/common-hints.ts

@@ -1,4 +1,5 @@
 import { HeuristicHint } from './heuristic-hint';
+import type { HeuristicHintInput } from './heuristic-hint';
 
 export const COMMON_HINTS = {
   add_to_cart: new HeuristicHint({
@@ -56,8 +57,20 @@ export const COMMON_HINTS = {
   }),
 } as const;
 
-export function getCommonHint(intent: string): HeuristicHint | null {
+/**
+ * Look up a heuristic hint by intent string.
+ *
+ * @param intent - The intent to look up (e.g., "add_to_cart", "book_flight")
+ * @param profileHints - Optional profile-provided hints to check after built-in hints
+ * @returns Matching HeuristicHint or null
+ */
+export function getCommonHint(
+  intent: string,
+  profileHints?: HeuristicHintInput[]
+): HeuristicHint | null {
   const normalized = intent.toLowerCase().replace(/[\s-]+/g, '_');
+
+  // Check built-in hints first
   const exactMatch = COMMON_HINTS[normalized as keyof typeof COMMON_HINTS];
   if (exactMatch) {
     return exactMatch;
@@ -69,5 +82,22 @@ export function getCommonHint(intent: string): HeuristicHint | null {
     }
   }
 
+  // Check profile-provided hints
+  if (profileHints && profileHints.length > 0) {
+    for (const ph of profileHints) {
+      const pattern = ph.intentPattern ?? ph.intent_pattern ?? '';
+      const phNormalized = pattern.toLowerCase().replace(/[\s-]+/g, '_');
+      if (normalized.includes(phNormalized) || phNormalized.includes(normalized)) {
+        return new HeuristicHint({
+          intentPattern: pattern,
+          textPatterns: ph.textPatterns ?? ph.text_patterns,
+          roleFilter: ph.roleFilter ?? ph.role_filter,
+          attributePatterns: ph.attributePatterns ?? ph.attribute_patterns,
+          priority: ph.priority,
+        });
+      }
+    }
+  }
+
   return null;
 }

src/agents/planner-executor/composable-heuristics.ts

@@ -3,23 +3,33 @@ import type { IntentHeuristics } from './planner-executor-agent';
 import { COMMON_HINTS } from './common-hints';
 import { HeuristicHint, type HeuristicHintInput } from './heuristic-hint';
 import { TaskCategory } from './task-category';
+import type { LearnedTargetFingerprint } from './profile-types';
 
 export interface ComposableHeuristicsOptions {
   staticHeuristics?: IntentHeuristics;
   taskCategory?: TaskCategory | null;
   useCommonHints?: boolean;
+  /** Learned fingerprints from previous successful runs */
+  learnedFingerprints?: LearnedTargetFingerprint[];
 }
 
+/** Minimum confidence for a learned fingerprint to be used as a hint */
+const MIN_FINGERPRINT_CONFIDENCE = 0.3;
+
 export class ComposableHeuristics implements IntentHeuristics {
   private readonly staticHeuristics: IntentHeuristics | null;
   private readonly taskCategory: TaskCategory | null;
   private readonly useCommonHints: boolean;
   private currentHints: HeuristicHint[] = [];
+  private readonly learnedFingerprints: LearnedTargetFingerprint[];
 
   constructor(options: ComposableHeuristicsOptions = {}) {
     this.staticHeuristics = options.staticHeuristics ?? null;
     this.taskCategory = options.taskCategory ?? null;
     this.useCommonHints = options.useCommonHints ?? true;
+    this.learnedFingerprints = (options.learnedFingerprints ?? []).filter(
+      fp => fp.confidence >= MIN_FINGERPRINT_CONFIDENCE
+    );
   }
 
   setStepHints(hints?: Array<HeuristicHint | HeuristicHintInput> | null): void {
@@ -48,6 +58,7 @@ export class ComposableHeuristics implements IntentHeuristics {
       return null;
     }
 
+    // 1. Check current step hints (highest priority)
     for (const hint of this.currentHints) {
       if (hint.matchesIntent(intent)) {
         const elementId = this.matchHint(hint, elements);
@@ -57,6 +68,13 @@ export class ComposableHeuristics implements IntentHeuristics {
       }
     }
 
+    // 2. Check learned fingerprints (dynamic hints from successful past runs)
+    const learnedMatch = this.matchLearnedFingerprint(intent, elements);
+    if (learnedMatch !== null) {
+      return learnedMatch;
+    }
+
+    // 3. Check common hints (built-in)
     if (this.useCommonHints) {
       const commonHint = this.getCommonHintForIntent(intent);
       if (commonHint) {
@@ -67,6 +85,7 @@ export class ComposableHeuristics implements IntentHeuristics {
       }
     }
 
+    // 4. Check static heuristics
     if (this.staticHeuristics) {
       try {
         const elementId = this.staticHeuristics.findElementForIntent(intent, elements, url, goal);
@@ -78,12 +97,14 @@ export class ComposableHeuristics implements IntentHeuristics {
       }
     }
 
+    // 5. Fall back to task category defaults
     return this.matchTaskCategoryDefaults(elements);
   }
 
   priorityOrder(): string[] {
     const patterns = [
       ...this.currentHints.map(hint => hint.intentPattern),
+      ...this.learnedFingerprints.map(fp => fp.intent),
       ...(this.useCommonHints ? Object.keys(COMMON_HINTS) : []),
     ];
 
@@ -108,6 +129,105 @@ export class ComposableHeuristics implements IntentHeuristics {
     return null;
   }
 
+  /**
+   * Match learned fingerprints against current snapshot elements.
+   * Fingerprints are sorted by confidence (descending) so the most
+   * reliable past success is tried first.
+   */
+  private matchLearnedFingerprint(intent: string, elements: SnapshotElement[]): number | null {
+    if (this.learnedFingerprints.length === 0) {
+      return null;
+    }
+
+    const normalizedIntent = intent.toLowerCase().replace(/[\s-]+/g, '_');
+    const sorted = [...this.learnedFingerprints].sort((a, b) => b.confidence - a.confidence);
+
+    for (const fp of sorted) {
+      // Match intent: exact or substring match
+      const fpIntent = fp.intent.toLowerCase().replace(/[\s-]+/g, '_');
+      if (
+        fpIntent !== normalizedIntent &&
+        !normalizedIntent.includes(fpIntent) &&
+        !fpIntent.includes(normalizedIntent)
+      ) {
+        continue;
+      }
+
+      for (const element of elements) {
+        if (this.fingerprintMatchesElement(fp, element)) {
+          return element.id;
+        }
+      }
+    }
+
+    return null;
+  }
+
+  /**
+   * Check if a learned fingerprint matches a snapshot element.
+   * Uses token overlap scoring with a minimum threshold.
+   */
+  private fingerprintMatchesElement(
+    fp: LearnedTargetFingerprint,
+    element: SnapshotElement
+  ): boolean {
+    let score = 0;
+    let maxScore = 0;
+
+    // Role match (weight: 2)
+    if (fp.role) {
+      maxScore += 2;
+      if ((element.role ?? '').toLowerCase() === fp.role) {
+        score += 2;
+      }
+    }
+
+    // Text token overlap (weight: up to 3)
+    if (fp.textTokens && fp.textTokens.length > 0) {
+      maxScore += 3;
+      const elementText = [element.text, element.ariaLabel, element.name]
+        .filter((v): v is string => typeof v === 'string')
+        .join(' ')
+        .toLowerCase();
+      const elementTokens = elementText.split(/\s+/).filter(t => t.length > 0);
+      const matchingTokens = fp.textTokens.filter(ft =>
+        elementTokens.some(et => et === ft || et.includes(ft))
+      );
+      if (matchingTokens.length > 0) {
+        score += Math.min(3, Math.ceil((matchingTokens.length / fp.textTokens.length) * 3));
+      }
+    }
+
+    // ARIA token overlap (weight: up to 2)
+    if (fp.ariaTokens && fp.ariaTokens.length > 0) {
+      maxScore += 2;
+      const ariaText = [element.ariaLabel, element.name]
+        .filter((v): v is string => typeof v === 'string')
+        .join(' ')
+        .toLowerCase();
+      const ariaTokens = ariaText.split(/\s+/).filter(t => t.length > 0);
+      const matchingTokens = fp.ariaTokens.filter(at =>
+        ariaTokens.some(et => et === at || et.includes(at))
+      );
+      if (matchingTokens.length > 0) {
+        score += Math.min(2, Math.ceil((matchingTokens.length / fp.ariaTokens.length) * 2));
+      }
+    }
+
+    // href path pattern match (weight: 2)
+    if (fp.hrefPathPattern) {
+      maxScore += 2;
+      const href = element.href || '';
+      if (href.toLowerCase().includes(fp.hrefPathPattern)) {
+        score += 2;
+      }
+    }
+
+    // Require at least 50% of available score to consider it a match
+    if (maxScore === 0) return false;
+    return score / maxScore >= 0.5;
+  }
+
   private getCommonHintForIntent(intent: string): HeuristicHint | null {
     const normalized = intent.toLowerCase().replace(/[\s-]+/g, '_');
     if (normalized in COMMON_HINTS) {