Fix AgentRuntime step_id from UUID to step-N

src/agent-runtime.ts

@@ -39,7 +39,6 @@
  */
 
 import { Page } from 'playwright';
-import { v4 as uuidv4 } from 'uuid';
 import { Snapshot } from './types';
 import { AssertContext, Predicate } from './verification';
 import { Tracer } from './tracing/tracer';
@@ -741,22 +740,22 @@ export class AgentRuntime {
    *
    * @param goal - Description of what this step aims to achieve
    * @param stepIndex - Optional explicit step index (otherwise auto-increments)
-   * @returns Generated stepId
+   * @returns Generated stepId in format 'step-N' where N is the step index
    */
   beginStep(goal: string, stepIndex?: number): string {
     // Clear previous step state
     this.assertionsThisStep = [];
 
-    // Generate new stepId
-    this.stepId = uuidv4();
-
     // Update step index
     if (stepIndex !== undefined) {
       this.stepIndex = stepIndex;
     } else {
       this.stepIndex += 1;
     }
 
+    // Generate stepId in 'step-N' format for Studio compatibility
+    this.stepId = `step-${this.stepIndex}`;
+
     return this.stepId;
   }
 

tests/actions.test.ts

@@ -174,7 +174,7 @@ describe('Actions', () => {
       } finally {
         await browser.close();
       }
-    }, 60000);
+    }, 90000); // 90 seconds - Windows CI can be slow
 
     it('should take snapshot after scroll when requested', async () => {
       const browser = await createTestBrowser();

tests/agent-runtime-assertions.test.ts

@@ -40,6 +40,52 @@ function makeElement(
   } as Element;
 }
 
+describe('AgentRuntime.beginStep() stepId format', () => {
+  it('generates stepId in step-N format', () => {
+    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);
+
+    const stepId1 = runtime.beginStep('Step 1');
+    expect(stepId1).toBe('step-1');
+    expect(runtime.stepIndex).toBe(1);
+
+    const stepId2 = runtime.beginStep('Step 2');
+    expect(stepId2).toBe('step-2');
+    expect(runtime.stepIndex).toBe(2);
+  });
+
+  it('generates stepId matching explicit stepIndex', () => {
+    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);
+
+    const stepId = runtime.beginStep('Custom step', 10);
+    expect(stepId).toBe('step-10');
+    expect(runtime.stepIndex).toBe(10);
+  });
+});
+
 describe('AgentRuntime.assert() with state predicates', () => {
   it('uses snapshot context for enabled/disabled/value assertions', () => {
     const sink = new MockSink();

tests/snapshot.test.ts

@@ -23,7 +23,7 @@ describe('Snapshot', () => {
     } finally {
       await browser.close();
     }
-  }, 60000); // 60 seconds - browser startup can be slow
+  }, 90000); // 90 seconds - Windows CI can be slow
 
   it('should have valid element structure', async () => {
     const browser = await createTestBrowser();

tests/video-recording.test.ts

@@ -301,5 +301,5 @@ describe('video recording', () => {
         throw error;
       }
     }
-  });
+  }, 180000); // 180 seconds - tests 3 resolutions with browser start/stop each, Windows CI can be slow
 });