tests: mcp agent evals for new tool

evals/mcp-eval-basic.config.json

@@ -121,6 +121,21 @@
         "browserbase_stagehand_get_url",
         "browserbase_session_close"
       ]
+    },
+    {
+      "name": "agent-basic-test",
+      "description": "Test autonomous agent functionality with simple task",
+      "steps": [
+        {
+          "user": "Create a browser session and use the agent to go to https://example.com and extract the page title, then close the session",
+          "expectedState": "Example Domain"
+        }
+      ],
+      "expectTools": [
+        "browserbase_session_create",
+        "browserbase_stagehand_agent",
+        "browserbase_session_close"
+      ]
     }
   ]
 }

evals/mcp-eval-minimal.config.json

@@ -67,6 +67,21 @@
         "browserbase_stagehand_get_url",
         "browserbase_session_close"
       ]
+    },
+    {
+      "name": "smoke-test-agent",
+      "description": "Quick test to verify autonomous agent works",
+      "steps": [
+        {
+          "user": "Create a browser session and use the agent to visit example.org and tell me the page title, then close the session",
+          "expectedState": "Example Domain"
+        }
+      ],
+      "expectTools": [
+        "browserbase_session_create",
+        "browserbase_stagehand_agent",
+        "browserbase_session_close"
+      ]
     }
   ]
 }

evals/mcp-eval.config.json

@@ -105,6 +105,21 @@
         "browserbase_session_create",
         "browserbase_stagehand_navigate"
       ]
+    },
+    {
+      "name": "agent-complex-task-test",
+      "description": "Test agent with multi-step task",
+      "steps": [
+        {
+          "user": "Create a browser session and use the agent to go to https://news.ycombinator.com, find the top post, and extract its title and score, then close the session",
+          "expectedState": "points"
+        }
+      ],
+      "expectTools": [
+        "browserbase_session_create",
+        "browserbase_stagehand_agent",
+        "browserbase_session_close"
+      ]
     }
   ]
 }

package.json

@@ -57,6 +57,7 @@
     "@changesets/cli": "^2.29.6",
     "@eslint/js": "^9.29.0",
     "@smithery/cli": "^1.2.15",
+    "@types/node": "^24.10.1",
     "chalk": "^5.3.0",
     "eslint": "^9.29.0",
     "eslint-plugin-react": "^7.37.5",