Test windows run examples workflow

.github/scripts/check_documented_examples.py

@@ -15,6 +15,11 @@
 from pathlib import Path
 
 
+EXEMPT_EXAMPLES = {
+    "examples/01_standalone_sdk/31_windows.py",
+}
+
+
 def find_documented_examples(docs_path: Path) -> set[str]:
     """
     Find all example file references in the docs repository.
@@ -149,6 +154,13 @@ def main() -> None:
     agent_examples = find_agent_sdk_examples(agent_sdk_root)
     print(f"   Found {len(agent_examples)} example file(s)")
 
+    exempt_present = agent_examples & EXEMPT_EXAMPLES
+    if exempt_present:
+        print("   Exempting the following example(s) from documentation check:")
+        for example in sorted(exempt_present):
+            print(f"     - {example}")
+        agent_examples -= exempt_present
+
     # Find all documented examples in docs
     print("\n📄 Scanning docs repository...")
     documented_examples = find_documented_examples(docs_path)

.github/workflows/run-examples.yml

@@ -20,15 +20,19 @@ permissions:
 
 jobs:
     test-examples:
-        if: github.event.label.name == 'test-examples' || github.event_name == 'workflow_dispatch' || github.event_name == 'schedule'
+        needs: test-examples-windows
+        if: ${{ always() && (github.event.label.name == 'test-examples' || contains(fromJson('["workflow_dispatch","schedule"]'), github.event_name)) 
+            }}
         runs-on: blacksmith-2vcpu-ubuntu-2404
         timeout-minutes: 60
         steps:
             - name: Wait for agent server to finish build
-              if: github.event_name == 'pull_request'
+              if: >
+                  github.event_name == 'pull_request' &&
+                  github.event.pull_request.head.repo.full_name == github.repository
               uses: lewagon/[email protected]
               with:
-                  ref: ${{ github.event.pull_request.head.ref }}
+                  ref: ${{ github.event.pull_request.head.sha }}
                   check-name: Build & Push (python-amd64)
                   repo-token: ${{ secrets.GITHUB_TOKEN }}
                   wait-interval: 10
@@ -49,6 +53,13 @@ jobs:
               with:
                   node-version: '22'
 
+            - name: Download Windows example results
+              if: needs.test-examples-windows.result == 'success'
+              uses: actions/download-artifact@v4
+              with:
+                  name: examples-results-windows
+                  path: .windows-example-results
+
             - name: Install dependencies
               run: uv sync --frozen --group dev
 
@@ -129,7 +140,13 @@ jobs:
                   uv run pytest tests/examples/test_examples.py \
                       --run-examples \
                       --examples-results-dir "$RESULTS_DIR" \
-                      -n 4 || EXIT_CODE=$?
+                      -n 4 \
+                      -k "not 31_windows.py" \
+                      || EXIT_CODE=$?
+
+                  if [ -d ".windows-example-results" ]; then
+                      cp .windows-example-results/*.json "$RESULTS_DIR"/ || true
+                  fi
 
                   TIMESTAMP="$(date -u '+%Y-%m-%d %H:%M:%S UTC')"
                   WORKFLOW_URL="${GITHUB_SERVER_URL}/${GITHUB_REPOSITORY}/actions/runs/${GITHUB_RUN_ID}"
@@ -178,3 +195,60 @@ jobs:
                       **Workflow Run:** ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
 
                       ${{ steps.read_report.outputs.report }}
+
+    test-examples-windows:
+        if: github.event.label.name == 'test-examples' || github.event_name == 'workflow_dispatch' || github.event_name == 'schedule'
+        runs-on: windows-latest
+        timeout-minutes: 60
+        steps:
+            - name: Checkout
+              uses: actions/checkout@v5
+              with:
+                  ref: ${{ github.event.pull_request.head.ref || github.ref }}
+                  repository: ${{ github.event.pull_request.head.repo.full_name || github.repository }}
+
+            - name: Install uv
+              uses: astral-sh/setup-uv@v7
+              with:
+                  enable-cache: true
+
+            - name: Install Node.js
+              uses: actions/setup-node@v4
+              with:
+                  node-version: '22'
+
+            - name: Install dependencies
+              run: uv sync --frozen --group dev
+
+            - name: Run Windows example
+              shell: bash
+              env:
+                  LLM_API_KEY: ${{ secrets.LLM_API_KEY }}
+                  LLM_MODEL: litellm_proxy/claude-haiku-4-5-20251001
+                  LLM_BASE_URL: https://llm-proxy.app.all-hands.dev
+                  RUNTIME_API_KEY: ${{ secrets.RUNTIME_API_KEY }}
+                  GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+                  PR_NUMBER: ${{ github.event.pull_request.number }}
+                  REPO_OWNER: ${{ github.repository_owner }}
+                  REPO_NAME: ${{ github.event.repository.name }}
+                  GITHUB_SHA: ${{ github.event.pull_request.head.sha }}
+              run: |
+                  RESULTS_DIR=".example-test-results"
+                  rm -rf "$RESULTS_DIR"
+                  mkdir -p "$RESULTS_DIR"
+
+                  uv run pytest tests/examples/test_examples.py \
+                      --run-examples \
+                      --examples-results-dir "$RESULTS_DIR" \
+                      -n auto \
+                      -k "31_windows.py"
+
+            - name: Upload Windows example results
+              if: always()
+              uses: actions/upload-artifact@v4
+              with:
+                  name: examples-results-windows
+                  path: .example-test-results
+                  include-hidden-files: true
+                  if-no-files-found: ignore
+

examples/01_standalone_sdk/30_tom_agent.py

@@ -90,6 +90,10 @@
 print("Tom agent consultation example completed!")
 print("=" * 80)
 
+# Report cost
+cost = llm.metrics.accumulated_cost
+print(f"EXAMPLE_COST: {cost}")
+
 
 # Optional: Index this conversation for Tom's user modeling
 # This builds user preferences and patterns from conversation history

examples/01_standalone_sdk/31_windows.py

@@ -0,0 +1,82 @@
+from __future__ import annotations
+
+import os
+import sys
+
+from pydantic import SecretStr
+
+from openhands.sdk import (
+    LLM,
+    Agent,
+    Conversation,
+    Event,
+    LLMConvertibleEvent,
+    get_logger,
+)
+from openhands.sdk.tool import Tool
+from openhands.tools.browser_use import BrowserToolSet
+from openhands.tools.file_editor import FileEditorTool
+
+
+logger = get_logger(__name__)
+
+api_key = os.getenv("LLM_API_KEY")
+if api_key is None:
+    raise RuntimeError("LLM_API_KEY environment variable is not set.")
+
+model = os.getenv("LLM_MODEL", "anthropic/claude-sonnet-4-5-20250929")
+base_url = os.getenv("LLM_BASE_URL")
+
+llm = LLM(
+    usage_id="agent",
+    model=model,
+    base_url=base_url,
+    api_key=SecretStr(api_key),
+)
+
+cwd = os.getcwd()
+tools = [
+    Tool(name=FileEditorTool.name),
+    Tool(name=BrowserToolSet.name),
+]
+
+agent = Agent(llm=llm, tools=tools)
+
+llm_messages = []
+
+
+def _safe_preview(text: str, limit: int = 200) -> str:
+    truncated = text[:limit]
+    encoding = getattr(sys.stdout, "encoding", None) or "utf-8"
+    return truncated.encode(encoding, errors="replace").decode(encoding)
+
+
+def conversation_callback(event: Event) -> None:
+    if isinstance(event, LLMConvertibleEvent):
+        llm_messages.append(event.to_llm_message())
+
+
+conversation = Conversation(
+    agent=agent,
+    callbacks=[conversation_callback],
+    workspace=cwd,
+    visualizer=None,
+)
+
+try:
+    conversation.send_message(
+        "Open https://openhands.dev/blog in the browser and summarize the key points "
+        "from the latest post."
+    )
+    conversation.run()
+finally:
+    conversation.close()
+
+print("=" * 100)
+print("Conversation finished. Got the following LLM messages:")
+for i, message in enumerate(llm_messages):
+    preview = _safe_preview(str(message))
+    print(f"Message {i}: {preview}")
+
+cost = llm.metrics.accumulated_cost
+print(f"EXAMPLE_COST: {cost}")

openhands-tools/openhands/tools/browser_use/definition.py

@@ -560,12 +560,26 @@ def create(
     ) -> list[ToolDefinition[BrowserAction, BrowserObservation]]:
         # Import executor only when actually needed to
         # avoid hanging during module import
-        from openhands.tools.browser_use.impl import BrowserToolExecutor
+        import sys
+
+        # Use Windows-specific executor on Windows systems
+        if sys.platform == "win32":
+            from openhands.tools.browser_use.impl_windows import (
+                WindowsBrowserToolExecutor,
+            )
+
+            executor = WindowsBrowserToolExecutor(
+                full_output_save_dir=conv_state.env_observation_persistence_dir,
+                **executor_config,
+            )
+        else:
+            from openhands.tools.browser_use.impl import BrowserToolExecutor
+
+            executor = BrowserToolExecutor(
+                full_output_save_dir=conv_state.env_observation_persistence_dir,
+                **executor_config,
+            )
 
-        executor = BrowserToolExecutor(
-            full_output_save_dir=conv_state.env_observation_persistence_dir,
-            **executor_config,
-        )
         # Each tool.create() returns a Sequence[Self], so we flatten the results
         tools: list[ToolDefinition[BrowserAction, BrowserObservation]] = []
         for tool_class in [