Adding the Integrations Tests

.github/workflows/agentex-tutorials-test.yml

@@ -0,0 +1,144 @@
+name: Test Tutorial Agents
+
+on:
+ workflow_dispatch:
+
+jobs:
+ find-tutorials:
+ runs-on: ubuntu-latest
+ outputs:
+ tutorials: ${{ steps.get-tutorials.outputs.tutorials }}
+ steps:
+ - name: Checkout agentex-python repo
+ uses: actions/checkout@v4
+
+ - name: Find all tutorials
+ id: get-tutorials
+ run: |
+ cd examples/tutorials
+ # Find all tutorials and exclude specific temporal ones
+ all_tutorials=$(find . -name "manifest.yaml" -exec dirname {} \; | sort | sed 's|^\./||')
+
+ # Filter out the specified temporal tutorials that are being updated
+ filtered_tutorials=$(echo "$all_tutorials" | grep -v -E "(temporal)")
+
+ # Convert to JSON array
+ tutorials=$(echo "$filtered_tutorials" | jq -R -s -c 'split("\n") | map(select(length > 0))')
+
+ echo "tutorials=$tutorials" >> $GITHUB_OUTPUT
+ echo "All tutorials found: $(echo "$all_tutorials" | wc -l)"
+ echo "Filtered tutorials: $(echo "$filtered_tutorials" | wc -l)"
+ echo "Excluded tutorials:"
+ echo "$all_tutorials" | grep -E "(10_temporal/050_|10_temporal/070_|10_temporal/080_)" || echo " (none matched exclusion pattern)"
+ echo "Final tutorial list: $tutorials"
+
+ test-tutorial:
+ needs: find-tutorials
+ runs-on: ubuntu-latest
+ timeout-minutes: 15
+ strategy:
+ matrix:
+ tutorial: ${{ fromJson(needs.find-tutorials.outputs.tutorials) }}
+ fail-fast: false
+ name: test-${{ matrix.tutorial }}
+
+ steps:
+ - name: Checkout agentex-python repo
+ uses: actions/checkout@v4
+
+ - name: Install UV
+ run: |
+ curl -LsSf https://astral.sh/uv/install.sh | sh
+ echo "$HOME/.local/bin" >> $GITHUB_PATH
+
+ - name: Checkout scale-agentex repo
+ uses: actions/checkout@v4
+ with:
+ repository: scaleapi/scale-agentex
+ path: scale-agentex
+
+ - name: Configure Docker Compose for host networking
+ run: |
+ cd scale-agentex/agentex
+ echo "🔧 Configuring AgentEx container for GitHub Actions networking..."
+
+ # Install yq for YAML manipulation
+ sudo wget -qO /usr/local/bin/yq https://github.com/mikefarah/yq/releases/latest/download/yq_linux_amd64
+ sudo chmod +x /usr/local/bin/yq
+
+ # Add extra_hosts to agentex service to make host.docker.internal work
+ yq eval '.services.agentex.extra_hosts = ["host.docker.internal:host-gateway"]' -i docker-compose.yml
+
+ echo "✅ Added extra_hosts configuration to agentex service"
+
+ - name: Start AgentEx Server
+ run: |
+ cd scale-agentex/agentex
+ echo "🚀 Starting AgentEx server and dependencies..."
+
+ # Start all services
+ docker compose up -d
+
+ echo "⏳ Waiting for dependencies to be healthy..."
+
+ # Wait for services to be healthy
+ for i in {1..30}; do
+ if docker compose ps | grep -q "healthy"; then
+ echo "✅ Dependencies are healthy"
+ break
+ fi
+ echo " Attempt $i/30: Waiting for services..."
+ sleep 5
+ done
+
+ # Wait specifically for AgentEx server to be ready
+ echo "⏳ Waiting for AgentEx server to be ready..."
+ for i in {1..30}; do
+ if curl -s --max-time 5 http://localhost:5003/health >/dev/null 2>&1; then
+ echo "✅ AgentEx server is ready"
+ break
+ fi
+ echo " Attempt $i/30: Waiting for AgentEx server..."
+ sleep 5
+ done
+
+ - name: Build AgentEx SDK
+ run: |
+ echo "🔨 Building AgentEx SDK wheel..."
+ uv build
+ echo "✅ SDK built successfully"
+ ls -la dist/
+
+ - name: Test Tutorial
+ working-directory: ./examples/tutorials
+ env:
+ OPENAI_API_KEY: ${{ secrets.TUTORIAL_OPENAI_API_KEY }}
+ HEALTH_CHECK_PORT: 8080 # Use non-privileged port for temporal worker health checks
+ run: |
+ echo "Testing tutorial: ${{ matrix.tutorial }}"
+ AGENTEX_API_BASE_URL="http://localhost:5003" \
+ ./run_agent_test.sh --build-cli "${{ matrix.tutorial }}"
+
+ - name: Record test result
+ id: test-result
+ if: always()
+ run: |
+ if [ "${{ steps.run-test.outcome }}" == "success" ]; then
+ echo "result=passed" >> $GITHUB_OUTPUT
+ else
+ echo "result=failed" >> $GITHUB_OUTPUT
+ fi
+
+ test-summary:
+ if: always()
+ needs: [find-tutorials, test-tutorial]
+ runs-on: ubuntu-latest
+ name: Test Summary
+ steps:
+ - name: Generate Test Summary
+ run: |
+ echo "# 🧪 Tutorial Tests Summary" >> $GITHUB_STEP_SUMMARY
+ echo "" >> $GITHUB_STEP_SUMMARY
+
+ # Get tutorial list from needs context
+ tutorials='${{ needs.find-tutorials.outputs.tutorials }}'

.github/workflows/build-and-push-tutorial-agent.yml

@@ -24,3 +24,91 @@ on:
  required: true
  type: string
  default: "latest"
+
+permissions:
+ contents: read
+ packages: write
+
+jobs:
+ build-and-push-agent:
+ timeout-minutes: 10
+ name: Build Tutorial Agent
+ runs-on: ubuntu-latest
+
+ steps:
+ - name: Checkout repository
+ uses: actions/checkout@v4
+
+ - name: Validate agent path exists
+ run: |
+ if [ ! -d "${{ inputs.agent_path }}" ]; then
+ echo "❌ Error: Agent path '${{ inputs.agent_path }}' does not exist"
+ exit 1
+ fi
+ echo "✅ Agent path verified: ${{ inputs.agent_path }}"
+
+ - name: Validate manifest.yaml exists
+ run: |
+ if [ ! -f "${{ inputs.agent_path }}/manifest.yaml" ]; then
+ echo "❌ Error: manifest.yaml not found in '${{ inputs.agent_path }}'"
+ exit 1
+ fi
+ echo "✅ manifest.yaml found"
+ echo "### Validation Summary" >> $GITHUB_STEP_SUMMARY
+ echo "- **Agent Path**: ${{ inputs.agent_path }}" >> $GITHUB_STEP_SUMMARY
+ echo "- **Version Tag**: ${{ inputs.version_tag }}" >> $GITHUB_STEP_SUMMARY
+ echo "- **Status**: ✅ Validation passed" >> $GITHUB_STEP_SUMMARY
+
+ - name: Set up Docker Buildx
+ uses: docker/setup-buildx-action@v3
+
+ - name: Set up Python
+ uses: actions/setup-python@v4
+ with:
+ python-version: "3.12"
+
+ - name: Get latest agentex-sdk version from PyPI
+ id: get-version
+ run: |
+ LATEST_VERSION=$(curl -s https://pypi.org/pypi/agentex-sdk/json | jq -r '.info.version')
+ echo "Latest agentex-sdk version: $LATEST_VERSION"
+ echo "AGENTEX_SDK_VERSION=$LATEST_VERSION" >> $GITHUB_ENV
+ pip install agentex-sdk==$LATEST_VERSION
+ echo "Installed agentex-sdk version $LATEST_VERSION"
+
+ - name: Generate Image name
+ id: image-name
+ run: |
+ # Remove examples/tutorials/ prefix and replace / with -
+ AGENT_NAME=$(echo "${{ inputs.agent_path }}" | sed 's|^examples/tutorials/||' | sed 's|/|-|g')
+ echo "AGENT_NAME=$AGENT_NAME" >> $GITHUB_ENV
+ echo "agent_name=$AGENT_NAME" >> $GITHUB_OUTPUT
+ echo "Agent name set to $AGENT_NAME"
+
+ - name: Login to GitHub Container Registry
+ uses: docker/login-action@v3
+ with:
+ registry: ghcr.io
+ username: ${{ github.actor }}
+ password: ${{ secrets.GITHUB_TOKEN }}
+
+ - name: Build and Push Agent Image
+ env:
+ REGISTRY: ghcr.io
+ run: |
+ AGENT_NAME="${{ steps.image-name.outputs.agent_name }}"
+ VERSION_TAG="${{ inputs.version_tag }}"
+ REPOSITORY_NAME="${{ github.repository }}/tutorial-agents/${AGENT_NAME}"
+ FULL_IMAGE="${REGISTRY}/${REPOSITORY_NAME}:${VERSION_TAG}"
+
+ agentex agents build \
+ --manifest "${{ inputs.agent_path }}/manifest.yaml" \
+ --registry "${REGISTRY}" \
+ --tag "${VERSION_TAG}" \
+ --platforms "linux/amd64" \
+ --repository-name "${REPOSITORY_NAME}" \
+ --push
+
+ echo "Successfully built and pushed: ${FULL_IMAGE}"
+ echo "### Build Complete" >> $GITHUB_STEP_SUMMARY
+ echo "- **Image**: \`${FULL_IMAGE}\`" >> $GITHUB_STEP_SUMMARY

examples/tutorials/10_agentic/00_base/040_other_sdks/project/acp.py

@@ -42,6 +42,7 @@
  config=AgenticACPConfig(type="base"),
 )
 
+
 class StateModel(BaseModel):
  input_list: List[dict]
  turn_number: int
@@ -53,11 +54,7 @@ class StateModel(BaseModel):
  args=["-y", "@modelcontextprotocol/server-sequential-thinking"],
  ),
  StdioServerParameters(
- command="uvx",
- args=["openai-websearch-mcp"],
- env={
- "OPENAI_API_KEY": os.environ.get("OPENAI_API_KEY", "")
- }
+ command="uvx", args=["openai-websearch-mcp"], env={"OPENAI_API_KEY": os.environ.get("OPENAI_API_KEY", "")}
  ),
 ]
 
@@ -72,6 +69,7 @@ async def handle_task_create(params: CreateTaskParams):
  )
  await adk.state.create(task_id=params.task.id, agent_id=params.agent.id, state=state)
 
+
 @acp.on_task_event_send
 async def handle_event_send(params: SendEventParams):
  # !!! Warning: Because "Agentic" ACPs are designed to be fully asynchronous, race conditions can occur if parallel events are sent. It is highly recommended to use the "temporal" type in the AgenticACPConfig instead to handle complex use cases. The "base" ACP is only designed to be used for simple use cases and for learning purposes.
@@ -85,7 +83,6 @@ async def handle_event_send(params: SendEventParams):
  if params.event.content.author != "user":
  raise ValueError(f"Expected user message, got {params.event.content.author}")
 
-
  # Retrieve the task state. Each event is handled as a new turn, so we need to get the state for the current turn.
  task_state = await adk.state.get_by_task_and_agent(task_id=params.task.id, agent_id=params.agent.id)
  if not task_state:
@@ -94,12 +91,8 @@ async def handle_event_send(params: SendEventParams):
  state.turn_number += 1
  # Add the new user message to the message history
  state.input_list.append({"role": "user", "content": params.event.content.content})
-
- async with adk.tracing.span(
- trace_id=params.task.id,
- name=f"Turn {state.turn_number}",
- input=state
- ) as span:
+
+ async with adk.tracing.span(trace_id=params.task.id, name=f"Turn {state.turn_number}", input=state) as span:
  # Echo back the user's message so it shows up in the UI. This is not done by default so the agent developer has full control over what is shown to the user.
  await adk.messages.create(
  task_id=params.task.id,
@@ -156,6 +149,7 @@ async def handle_event_send(params: SendEventParams):
  if span:
  span.output = state
 
+
 @acp.on_task_cancel
 async def handle_task_cancel(params: CancelTaskParams):
  """Default task cancel handler"""
@@ -173,8 +167,8 @@ async def mcp_server_context(mcp_server_params: list[StdioServerParameters]):
  servers = []
  for params in mcp_server_params:
  server = MCPServerStdio(
- name=f"Server: {params.command}", 
- params=params.model_dump(), 
+ name=f"Server: {params.command}",
+ params=params.model_dump(),
  cache_tools_list=True,
  client_session_timeout_seconds=60,
  )
@@ -253,7 +247,6 @@ async def run_openai_agent_with_custom_streaming(
  try:
  # Process streaming events with TaskMessage creation
  async for event in result.stream_events():
-
  if event.type == "run_item_stream_event":
  if event.item.type == "tool_call_item":
  tool_call_item = event.item.raw_item
@@ -374,9 +367,7 @@ async def run_openai_agent_with_custom_streaming(
  if span:
  span.output = {
  "new_items": [
- item.raw_item.model_dump()
- if isinstance(item.raw_item, BaseModel)
- else item.raw_item
+ item.raw_item.model_dump() if isinstance(item.raw_item, BaseModel) else item.raw_item
  for item in result.new_items
  ],
  "final_output": result.final_output,

examples/tutorials/10_agentic/00_base/040_other_sdks/tests/test_agent.py

@@ -114,7 +114,7 @@ async def test_send_event_and_poll_simple_query(self, client: AsyncAgentex, agen
  break
 
  # Verify state has been updated by polling the states for 10 seconds
- for i in range(10):
+ for i in range(20):
  if i == 9:
  raise Exception("Timeout waiting for state updates")
  states = await client.states.list(agent_id=agent_id, task_id=task.id)
@@ -187,7 +187,12 @@ async def test_multi_turn_conversation_with_state(self, client: AsyncAgentex, ag
  sleep_interval=1.0,
  ):
  assert isinstance(message, TaskMessage)
- if message.content and message.content.type == "text" and message.content.author == "agent" and message.content.content:
+ if (
+ message.content
+ and message.content.type == "text"
+ and message.content.author == "agent"
+ and message.content.content
+ ):
  break
 
  ## keep polling the states for 10 seconds for the input_list and turn_number to be updated
@@ -216,7 +221,12 @@ async def test_multi_turn_conversation_with_state(self, client: AsyncAgentex, ag
  timeout=30,
  sleep_interval=1.0,
  ):
- if message.content and message.content.type == "text" and message.content.author == "agent" and message.content.content:
+ if (
+ message.content
+ and message.content.type == "text"
+ and message.content.author == "agent"
+ and message.content.content
+ ):
  response_text = message.content.content.lower()
  assert "blue" in response_text
  found_response = True
@@ -273,7 +283,10 @@ async def stream_messages() -> None:
  # For full messages, content is at the top level
  # For delta messages, we need to check parent_task_message
  if msg_type == "full":
- if event.get("content", {}).get("type") == "text" and event.get("content", {}).get("author") == "user":
+ if (
+ event.get("content", {}).get("type") == "text"
+ and event.get("content", {}).get("author") == "user"
+ ):
  user_message_found = True
  elif msg_type == "done":
  break