Yeesian Ng

This commit delivers a **bug fix** to the **Vertex AI client initialization logic** within the `Vertex AI Memory Bank Service` and `Vertex AI Session Service`. It modifies the `_get_api_client` method in `src/google/adk/memory/vertex_ai_memory_bank_service.py` and `src/google/adk/sessions/vertex_ai_session_service.py` to conditionally initialize the Vertex AI client. This change ensures that the client is configured with either project/location *or* an API key, but not both, resolving the `Project/location and API key are mutually exclusive` error. The fix prevents runtime failures and ensures correct API client setup for both ExpressMode and standard GCP project environments.

This commit introduces a **new capability** to the **AgentEngine deployment** process, enabling users to specify the Google Cloud region through the `GOOGLE_CLOUD_LOCATION` environment variable. Specifically, the `to_agent_engine` function within the `src/google/adk/cli/cli_deploy.py` module has been updated to read and utilize this environment variable. This enhancement streamlines the deployment workflow by providing an alternative, environment-driven method for configuring the target region, reducing the need for explicit command-line arguments. The change primarily impacts users deploying agents to **Google Cloud AgentEngine**, offering greater flexibility in their deployment scripts.

This commit **adds back unit tests** for the **CLI utility responsible for deploying to AgentEngine**, specifically verifying the happy path of the `to_agent_engine` function in `src/google/adk/cli/cli_deploy.py`. Concurrently, it **refactors** the **AgentEngine deployment logic** to handle working directory changes and ensure correct dependency inclusion. The `staging_bucket` argument and CLI option are **deprecated** and removed from the `cli_deploy_agent_engine` command, streamlining the **CLI deployment interface**. This work **improves test coverage** and cleans up the **CLI deployment interface** for AgentEngine.

This commit provides a **bug fix** to ensure the correct deployment of **config-based agents** to `AgentEngine`. Previously, an incorrect path resolution within the **ADK CLI's deployment logic** prevented these agents from being properly located during the deployment process. The fix involves simplifying the agent folder path and removing an unnecessary `FileNotFoundError` handler in `src/google/adk/cli/cli_deploy.py`. This change guarantees that **config-based agents** are now correctly identified and deployed, improving the reliability of the **AgentEngine deployment process**.

This commit provides a **bug fix** and **refactoring** to the **`adk` CLI deployment process**. It specifically addresses how **requirements files are handled and prioritized** when deploying to the **agent engine**. The logic within the `to_agent_engine` function in `src/google/adk/cli/cli_deploy.py` has been updated to correctly overwrite existing requirement files and create default ones if none are specified. This ensures robust dependency management and improves the reliability and consistency of agent engine deployments.

This commit introduces a **new capability** to enhance logging during **Agent Engine deployment**. It configures the `vertexai_genai.agentengines` logger to output messages at the `INFO` level when the `cli_deploy_agent_engine` command is executed within the `cli_tools_click.py` module. This change improves observability by providing more detailed feedback to users during the deployment process, making it easier to monitor and troubleshoot.

This commit **fixes** an issue by integrating **Vertex AI initialization** into the **agent engine app template**. It modifies `src/google/adk/cli/cli_deploy.py` to ensure that deployments to **AgentEngine** in non-express mode correctly initialize Vertex AI with the specified project and location. This **maintenance fix** is crucial for the proper functioning of deployed agent code that relies on Vertex AI services.

This commit **updates the project's dependency version constraints** to ensure they are accurately based on versions published on PyPI. This **bug fix** modifies the system responsible for defining external package requirements, aligning them with publicly available releases. The change primarily affects the **dependency management** and **build configuration** subsystems, enhancing **installation reliability** and **compatibility** by preventing discrepancies between internal and external versioning.

This commit introduces a **new capability** by adding **support for Vertex AI Express Mode** when deploying applications to the **Agent Engine**. It updates the `adk deploy agent_engine` command, primarily in `src/google/adk/cli/cli_deploy.py` and `src/google/adk/cli/cli_tools_click.py`, to include new command-line options for configuring API keys and flexible app object naming. This enhancement streamlines the deployment workflow, allowing users to leverage Vertex AI Express Mode's benefits for their Agent Engine applications. The change significantly expands the deployment options available, providing greater flexibility in application configuration and API key management for users.

This commit introduces a **new feature** to the **ADK CLI deployment process**, specifically for **agent engines**. It adds a `--agent_engine_config_file` option to the `adk deploy agent_engine` command, allowing users to specify a JSON file for agent deployment configuration. This configuration file is processed by the `to_agent_engine` function, enabling the loading of default settings which can then be overridden by other command-line arguments. This enhancement provides greater flexibility and standardization for configuring agent engine deployments.

This commit introduces a **bug fix** to the **CLI deployment process** for `google-adk` agents. It ensures that the `agent_framework` parameter is correctly set to `'google-adk'` during the initialization of the `AgentEngine` within the `to_agent_engine` function in `src/google/adk/cli/cli_deploy.py`. This change is crucial for the proper identification and operation of **module-based agent engines**, preventing potential issues arising from an unset or incorrect framework identifier. It directly impacts how `google-adk` agents are deployed and recognized by the system.

This commit introduces a **new feature** to the **Agent Engine deployment process**, enabling users to explicitly exclude files and directories from being deployed. It implements support for a new `.ae_ignore` file within the **CLI deployment tool** (`cli_deploy.py`), which is now respected by the `to_agent_engine` function. This enhancement prevents unnecessary or sensitive files from being uploaded, thereby improving deployment efficiency and security. Additionally, the commit includes a minor adjustment to the `.env` file lookup path during deployment.

This commit **fixes** the **ADK CLI deployment process** by switching from using `agent_engine_name` to `agent_engine_id` for identifying and updating agent engine instances. The `cli_deploy_agent_engine` command in `src/google/adk/cli/cli_tools_click.py` now expects the `--agent_engine_id` flag instead of `--agent_engine_name`, with corresponding help text updates. Internally, the `to_agent_engine` function in `src/google/adk/cli/cli_deploy.py` has been refactored to utilize this new identifier, along with updates to module and system path handling. This change ensures more robust and reliable identification of agent engine instances within the **ADK CLI**.

This commit introduces a **new feature** to the **CLI deployment tools**, enabling users to **update existing agent engine instances** rather than just creating new ones. It modifies the `cli_deploy_agent_engine` command in `src/google/adk/cli/cli_tools_click.py` by adding a new `--agent_engine_name` option. This option allows specifying an existing agent engine, which the `to_agent_engine` function in `src/google/adk/cli/cli_deploy.py` then uses to perform an update operation. This enhancement significantly improves the **management capabilities** for agent engines, allowing for iterative changes and maintenance directly through the command line.

This commit introduces a **new deployment option** to **fix potential import resolution issues** when deploying to the **agent engine** via the **ADK CLI**. A new `--absolutize_imports` command-line option is added to `cli_deploy_agent_engine`, which is then passed to the `to_agent_engine` function. This enables the conversion of relative imports to absolute imports during the deployment process, ensuring correct module loading in the target environment. The change also includes a **refactoring** of temporary directory handling within the deployment logic.

This commit **fixes** an issue in the `adk` CLI where project and region settings were not properly prioritized during **deployment to Agent Engine**. It updates the `to_agent_engine` function in `src/google/adk/cli/cli_deploy.py` to ensure that command-line arguments for `--project` and `--region` correctly **override** any corresponding values found in the `.env` file. Additionally, the help messages for these CLI options in `src/google/adk/cli/cli_tools_click.py` have been updated to clarify this precedence. This **bug fix** improves user understanding and prevents configuration errors, ensuring consistent and predictable configuration behavior for Agent Engine deployments.

This commit introduces a **new capability** to the **CLI agent deployment process**, enabling users to specify a `display_name` and `description` for agents. The `to_agent_engine` function in `src/google/adk/cli/cli_deploy.py` has been updated to accept these optional arguments, enhancing the metadata associated with deployed agents. Correspondingly, new `--display_name` and `--description` **CLI options** have been added to the `cli_deploy_agent_engine` command in `src/google/adk/cli/cli_tools_click.py`. This allows users to provide richer, more descriptive information when deploying agents to the **agent engine**, improving their discoverability and management.

This commit introduces a **new feature** to the **ADK CLI**, enabling users to deploy agents directly to **Vertex AI Agent Engine**. It extends the **ADK CLI's deployment capabilities** by adding a new `agent_engine` command under `adk deploy` in `cli_tools_click.py`, providing a dedicated user interface for this new target. The underlying logic for handling the deployment to Vertex AI Agent Engine is implemented in the new `to_agent_engine` function within `cli_deploy.py`. This enhancement significantly expands the deployment options available within the **ADK CLI**, streamlining the process for users targeting Vertex AI.