mirror Python recaptcha heuristics like selector hits

Author: SentienceDEV

src/agent-runtime.ts

@@ -863,9 +863,43 @@ export class AgentRuntime {
     const iframeHits = evidence?.iframe_src_hits ?? [];
     const urlHits = evidence?.url_hits ?? [];
     const textHits = evidence?.text_hits ?? [];
+    const selectorHits = evidence?.selector_hits ?? [];
     if (iframeHits.length === 0 && urlHits.length === 0 && textHits.length === 0) {
       return false;
     }
+    // Heuristic: many sites include passive reCAPTCHA badges (v3) that should not block.
+    // Only block when there is evidence of an interactive challenge.
+    const hitsAll = [...iframeHits, ...urlHits, ...textHits, ...selectorHits];
+    const hitsLower = hitsAll.map(hit => String(hit || '').toLowerCase()).filter(Boolean);
+    const joinedHits = hitsLower.join(' ');
+    const strongText = [
+      "i'm not a robot",
+      'verify you are human',
+      'human verification',
+      'complete the security check',
+      'please verify',
+    ].some(needle => joinedHits.includes(needle));
+    const strongIframe = hitsLower.some(hit =>
+      ['api2/bframe', 'hcaptcha', 'turnstile'].some(needle => hit.includes(needle))
+    );
+    const strongSelector = hitsLower.some(hit =>
+      [
+        'g-recaptcha-response',
+        'h-captcha-response',
+        'cf-turnstile-response',
+        'recaptcha-checkbox',
+        'hcaptcha-checkbox',
+      ].some(needle => hit.includes(needle))
+    );
+    const onlyGeneric =
+      !strongText &&
+      !strongIframe &&
+      !strongSelector &&
+      hitsLower.length > 0 &&
+      hitsLower.every(hit => hit.includes('captcha') || hit.includes('recaptcha'));
+    if (onlyGeneric) {
+      return false;
+    }
     const confidence = captcha.confidence ?? 0;
     const minConfidence = options.minConfidence ?? DEFAULT_CAPTCHA_OPTIONS.minConfidence;
     return confidence >= minConfidence;

tests/agent-runtime-captcha-detection.test.ts

@@ -0,0 +1,83 @@
+import { AgentRuntime } from '../src/agent-runtime';
+import { TraceSink } from '../src/tracing/sink';
+import { Tracer } from '../src/tracing/tracer';
+import { CaptchaDiagnostics, Snapshot } from '../src/types';
+import { MockPage } from './mocks/browser-mock';
+
+class MockSink extends TraceSink {
+  public events: any[] = [];
+  emit(event: Record<string, any>): void {
+    this.events.push(event);
+  }
+  async close(): Promise<void> {
+    // no-op
+  }
+  getSinkType(): string {
+    return 'MockSink';
+  }
+}
+
+function makeRuntime(): AgentRuntime {
+  const sink = new MockSink();
+  const tracer = new Tracer('test-run', sink);
+  const page = new MockPage('https://example.com') as any;
+  const browserLike = {
+    snapshot: async () => ({
+      status: 'success',
+      url: 'https://example.com',
+      elements: [],
+      timestamp: 't1',
+    }),
+  };
+  const runtime = new AgentRuntime(browserLike as any, page as any, tracer);
+  runtime.setCaptchaOptions({ minConfidence: 0.1, policy: 'abort' });
+  return runtime;
+}
+
+function makeSnapshot(captcha: CaptchaDiagnostics): Snapshot {
+  return {
+    status: 'success',
+    url: 'https://example.com',
+    elements: [],
+    diagnostics: { captcha },
+    timestamp: 't1',
+  };
+}
+
+describe('AgentRuntime captcha detection', () => {
+  it('ignores passive recaptcha badges', () => {
+    const runtime = makeRuntime();
+    const captcha: CaptchaDiagnostics = {
+      detected: true,
+      confidence: 0.95,
+      provider_hint: 'recaptcha',
+      evidence: {
+        iframe_src_hits: ['https://www.google.com/recaptcha/api2/anchor?ar=1'],
+        selector_hits: [],
+        text_hits: [],
+        url_hits: [],
+      },
+    };
+
+    const detected = (runtime as any).isCaptchaDetected(makeSnapshot(captcha));
+    expect(detected).toBe(false);
+  });
+
+  it('detects interactive captcha challenges', () => {
+    const runtime = makeRuntime();
+    const captcha: CaptchaDiagnostics = {
+      detected: true,
+      confidence: 0.95,
+      provider_hint: 'recaptcha',
+      evidence: {
+        iframe_src_hits: [],
+        selector_hits: [],
+        text_hits: ["I'm not a robot"],
+        url_hits: [],
+      },
+    };
+
+    const detected = (runtime as any).isCaptchaDetected(makeSnapshot(captcha));
+    expect(detected).toBe(true);
+  });
+});