This change will help the tools user identify per agent job usage in BQ console and INFORMATION_SCHEMA views. This change fulfills the feature request #3582. Here is a demo after change: screen/C6YB4ge2FM2ZREi.
PiperOrigin-RevId: 834480140
When this agent was built before the agent name was changed in the build script and accepted like that from the VB so changed to set into that form
Co-authored-by: George Weale <gweale@google.com>
PiperOrigin-RevId: 834479335
This change introduces `input_transcription` and `output_transcription` fields to session events, enabling the storage and retrieval of transcription data in both the database and Vertex AI session services.
Closes#3172
Co-authored-by: Hangfei Lin <hangfei@google.com>
PiperOrigin-RevId: 834366848
Fixes https://github.com/google/adk-python/issues/3174
The event compaction process, configured via `EventsCompactionConfig`, was
previously scheduled as a background task using `asyncio.create_task`.
Because Python's `asyncio.create_task` only holds a weak reference to
the created task and no strong reference was maintained by ADK, the
compaction task could be garbage-collected before it finished executing.
This resulted in event compaction failing silently or only partially
running, preventing session history from being summarized.
### Approaches Considered
Two approaches were considered to fix this:
1. **`asyncio.create_task` + Reference:** Create the task with `create_task` and store a strong reference to it (e.g., in a `set` on the [Runner](http://_vscodecontentref_/0) instance), removing it only when complete via `task.add_done_callback()`.
* **Pros:** The [run_async](http://_vscodecontentref_/1) async generator finishes immediately after yielding the last agent event.
* **Cons:** Adds complexity to the Runner state; background task failures are silent to the [run_async](http://_vscodecontentref_/2) caller; requires enhancement to `runner.close()` to correctly manage pending tasks on shutdown.
2. **`await`:** Directly `await` the compaction coroutine at the end of [run_async](http://_vscodecontentref_/3) after all agent events have been yielded.
* **Pros:** Simple to implement; ensures compaction runs to completion; failures during compaction propagate immediately to the [run_async](http://_vscodecontentref_/4) caller, making them visible and easier to debug.
* **Cons:** The `async for` loop iterating over [run_async](http://_vscodecontentref_/5) will not terminate until compaction finishes.
### Decision
This change implements the `await` approach. Although it means the `async for` loop takes longer to terminate when compaction occurs, it was chosen for its **simplicity and robustness**. Ensuring that compaction either succeeds or fails visibly is preferable to silent background failures.
All agent response events are yielded *before* compaction starts, so there is **no user-perceived delay in receiving the agent's answer** for the current turn.
### Integration Note for Users
Because compaction is now awaited, code consuming events via `async for event in runner.run_async(...)` will only finish iterating *after* compaction is complete (if compaction is triggered for that invocation).
If your application only needs the agent's response to proceed (e.g., displaying a message in a UI and allowing the user to reply), you can process events as they arrive and initiate the next turn without waiting for the `async for` loop to fully terminate. A new call to [run_async](http://_vscodecontentref_/6) for the next user query can be made immediately and will execute concurrently.
**Example:**
```python
async def handle_agent_turn(runner, message):
print("Agent is thinking...")
async for event in runner.run_async(user_id='...', session_id='...', new_message=message):
# Stream events to UI, log, etc.
if event.author == 'model' and event.content and event.content.parts[0].text:
print(f"Agent response: {event.content.parts[0].text}")
# The agent has provided a text response.
# The application can now enable user input for the next turn,
# even though this async for loop might not finish immediately
# if compaction is running.
print("Invocation complete (including compaction if any).")
# In your application:
# A new call to handle_agent_turn(runner, next_message) can be made
# as soon as the user provides the next input, without waiting for
# the previous call's generator to be exhausted.
Co-authored-by: Hangfei Lin <hangfei@google.com>
PiperOrigin-RevId: 833885375
this creates service_factory to handle .adk folder changes (including per-agent .adk defaults and in-memory/custom URI handling)
Co-authored-by: George Weale <gweale@google.com>
PiperOrigin-RevId: 833875524
This change loads the agent or app from the specified directory before creating the session. This allows using the correct application name (from the `App` object if applicable) when initializing the session, rather than always defaulting to the folder name. The variable `root_agent` is also renamed to `agent_or_app` to better reflect that it can be either an Agent or an App
Co-authored-by: George Weale <gweale@google.com>
PiperOrigin-RevId: 833839070
Currently logic to check for a disconnected session only checks for certain headers but doesn't detect all cases, leading to situations where it tries to connect to a session that is down. This adds logic so that we ping the server to check if it is disconnected. Fixes https://github.com/google/adk-python/issues/3321.
Co-authored-by: Kathy Wu <wukathy@google.com>
PiperOrigin-RevId: 833487825
Updates type annotations to use built-in types like `list` and `dict`, and uses `| None` for optional types, along with adding `from __future__ import annotations`
Co-authored-by: George Weale <gweale@google.com>
PiperOrigin-RevId: 832522395
Currently logic to check for a disconnected session only checks for certain headers but doesn't detect all cases, leading to situations where it tries to connect to a session that is down. This adds logic so that we ping the server to check if it is disconnected. Fixes https://github.com/google/adk-python/issues/3321.
Co-authored-by: Kathy Wu <wukathy@google.com>
PiperOrigin-RevId: 832460068
This update enhances the BigQuery agent analytics plugin:
* **Schema Field Descriptions:** The recommended BigQuery table schema now includes descriptions for each field, improving data understandability.
* **Optimized Table Structure:** The plugin now creates the table with daily partitioning on `timestamp` and clustering on `event_type`, `agent`, and `user_id` by default.
* **Truncation Flag:** A new boolean field `is_truncated` is added to the schema to show if the `content` was truncated.
PiperOrigin-RevId: 832436799
In order to make evals more resilient to temporary model failures, we add retry options to llm_requests that are made during evals.
Note that this is a fall back option, if the developer has already specified their own retry options, then those will be honored.
Co-authored-by: Ankur Sharma <ankusharma@google.com>
PiperOrigin-RevId: 832383755
Two tools - detect_anomalies and analyze_contribution are modifying the settings passed to them, which is not right as the settings are held and passed by the top level, which means several tools share the same settings.
PiperOrigin-RevId: 832081738
This is ~27% of the current cold start latency after previous changes are submitted. This change refactors `google.adk.tools/__init__.py` to use `__getattr__` for lazy loading of all tools and related classes. Previously, all modules were imported directly upon `google.adk.tools` import, leading to potentially long initial import times. With lazy loading, modules are only imported when they are first accessed, improving the initial startup performance.
Co-authored-by: Liang Wu <wuliang@google.com>
PiperOrigin-RevId: 831537187
This change:
* Deprecates the `save_live_audio` configuration option in `RunConfig`, introducing `save_live_blob` as its replacement. A warning is issued if `save_live_audio` is used.
* Updates `base_llm_flow.py` to use the new `save_live_blob` flag.
* Ensures that audio events generated by `audio_cache_manager.py` are properly appended to the session service.
* Adds a utility script `pcm_audio_player.py` for playing raw PCM audio files.
Input sample event: 14a5859f-6b6c-46ed-9f28-e5008793b1c6|live_bidi_streaming_multi_agent|user|1e867c7f-dbe1-4268-a7bc-7a9fa5fbd16c|e-7a28a060-29bf-4483-bc5c-17248698a897|1762916981.5932|{"content":{"parts":[{"file_data":{"file_uri":"artifact://live_bidi_streaming_multi_agent/user/1e867c7f-dbe1-4268-a7bc-7a9fa5fbd16c/_adk_live/adk_live_audio_storage_input_audio_1762916981593.pcm#0","mime_type":"audio/pcm"}}],"role":"user"},"invocation_id":"e-7a28a060-29bf-4483-bc5c-17248698a897","author":"user","actions":{"state_delta":{},"artifact_delta":{},"requested_auth_configs":{},"requested_tool_confirmations":{}},"id":"14a5859f-6b6c-46ed-9f28-e5008793b1c6","timestamp":1762916981.5932002}
output sample event:
506c9df4-e143-4ebc-90a5-2f5b2eb26754|live_bidi_streaming_multi_agent|user|1e867c7f-dbe1-4268-a7bc-7a9fa5fbd16c|e-7a28a060-29bf-4483-bc5c-17248698a897|1762916986.10579|{"content":{"parts":[{"file_data":{"file_uri":"artifact://live_bidi_streaming_multi_agent/user/1e867c7f-dbe1-4268-a7bc-7a9fa5fbd16c/_adk_live/adk_live_audio_storage_output_audio_1762916986105.pcm;rate=24000#0","mime_type":"audio/pcm;rate=24000"}}],"role":"model"},"invocation_id":"e-7a28a060-29bf-4483-bc5c-17248698a897","author":"model","actions":{"state_delta":{},"artifact_delta":{},"requested_auth_configs":{},"requested_tool_confirmations":{}},"id":"506c9df4-e143-4ebc-90a5-2f5b2eb26754","timestamp":1762916986.105794}
Co-authored-by: Hangfei Lin <hangfei@google.com>
PiperOrigin-RevId: 831512074
pin crew ai to 3.13 at highest version as it uses chromadb and that uses onnxruntime which does not work yet with 3.14
Co-authored-by: George Weale <gweale@google.com>
PiperOrigin-RevId: 831508884
The change adds an extension point for controlling which request metadata gets attached to A2A requests made by a RemoteAgent.
Instead of taking metadata from custom_metadata of session events users can construct payloads using a2a_request_meta_provider.
request_meta feature was added in v0.3.11 of the a2a-sdk library: https://github.com/a2aproject/a2a-python/releases/tag/v0.3.11
PiperOrigin-RevId: 831506364
This CL adds debug logs to show the history being sent to the live connection and the live connect config used.
Co-authored-by: Hangfei Lin <hangfei@google.com>
PiperOrigin-RevId: 831473597
This changes how content parts are converted for LiteLLM, treating all "text/" mime types as plain text and only "application/pdf" as a file
Close#1940
Co-authored-by: George Weale <gweale@google.com>
PiperOrigin-RevId: 831418696
This is targeting the Client import mostly, but also prevents future latency increase if the other modules in genai adds more 3p dependencies. This change will make ADK only import `live`, `Client` and `_transformers` just-in-time, therefore cutting down cold start latency.
PiperOrigin-RevId: 831244349
This update improves the `BigQueryAgentAnalyticsPlugin` in several ways:
* Corrects the PyArrow schema generation to accurately reflect BigQuery field nullability based on the `mode` attribute.
* Introduces a configurable `shutdown_timeout` in `BigQueryLoggerConfig` to manage how long the plugin waits for pending logs to flush during shutdown.
* Adds more robust error handling within the `shutdown` method and background write tasks, particularly for event loop closure issues.
* Improves internal logging to provide better diagnostics.
* Ensures consistent use of safe content formatting.
PiperOrigin-RevId: 831225837
This change introduces a mechanism for users to register their own custom backend services for sessions, memory, and artifacts without modifying the ADK framework. This enhances the extensibility of ADK.
Two methods of registration are supported, both by placing a file in the parent directory of the agents.
**YAML Configuration (services.yaml or .yml)**
This is the recommended approach for simple services that can be instantiated with a constructor like MyService(uri="...", **kwargs).
Example services.yaml:
```
services:
- scheme: mysession
type: session
class: my_package.my_module.MyCustomSessionService
```
**Python Registration (services.py)**
For services requiring more complex initialization logic, users can define factory functions in a services.py file.
Example services.py
```
from google.adk.cli.service_registry import get_service_registry
from my_package.my_module import MyCustomSessionService
def my_session_factory(uri: str, **kwargs):
# custom initialization logic
return MyCustomSessionService(...)
get_service_registry().register_session_service("mysession", my_session_factory)
```
ADK will load services from services.yaml/.yml first, and then from services.py. If the same service scheme is defined in both, the registration in services.py will take precedence.
To use a registered service, specify its URI via the corresponding command-line flag, e.g., `--session_service_uri=mysession://....`
Co-authored-by: Shangjie Chen <deanchen@google.com>
PiperOrigin-RevId: 831211371
Creates AdkFolderManager for creating/resetting the .adk layout, helper builders that return SQLite- and filesystembacked services for each agent
Co-authored-by: George Weale <gweale@google.com>
PiperOrigin-RevId: 831206377
Retrying only on closed_resource error is not enough to be reliable for production environments due to the other network errors that may occur -- remote protocol error, read timeout, etc. We will update this to retry on all errors. Since it is only a one-time retry, it should not affect latency significantly. Fixes https://github.com/google/adk-python/issues/2561.
PiperOrigin-RevId: 831153803
The `google.cloud.storage` module is now imported within the `__init__` method of `GCSArtifactService`, rather than at the top level. This avoids importing the potentially heavy `storage` module unless an instance of `GCSArtifactService` is actually created.
Co-authored-by: Liang Wu <wuliang@google.com>
PiperOrigin-RevId: 831124463
This is targeting the Client import mostly, but also prevents future latency increase if the other modules in genai adds more 3p dependencies. This change will make ADK only import `live`, `Client` and `_transformers` just-in-time, therefore cutting down cold start latency.
Co-authored-by: Liang Wu <wuliang@google.com>
PiperOrigin-RevId: 831124329
Retrying only on closed_resource error is not enough to be reliable for production environments due to the other network errors that may occur -- remote protocol error, read timeout, etc. We will update this to retry on all errors. Since it is only a one-time retry, it should not affect latency significantly. Fixes https://github.com/google/adk-python/issues/2561.
Co-authored-by: Kathy Wu <wukathy@google.com>
PiperOrigin-RevId: 831123085
This change updates `RemoteA2AAgent` to extract and forward custom metadata from session events to the `a2a-sdk`'s `send_message` method. The metadata is looked for under the key `A2A_METADATA_PREFIX + "metadata"` within the `custom_metadata` of the relevant session events. The `a2a-sdk` dependency is also updated to a version that supports this feature.
This feature was added in v0.3.11 of the a2a-sdk library: https://github.com/a2aproject/a2a-python/releases/tag/v0.3.11
PiperOrigin-RevId: 831120978
This change introduces a shutdown lifecycle hook for plugins. The `PluginManager` now has an `async def shutdown()` method that will call `await plugin.shutdown()` on any registered plugins that implement the method. This is called from `Runner.close()`, allowing plugins to perform cleanup tasks like flushing logs or closing connections when the runner instance is being closed. This improves the reliability of plugins that perform background operations.
PiperOrigin-RevId: 831037737
The agent's instructions are updated to guide it to guess a positive integer, starting from 50 and decreasing, using the `guess_number_tool` until the target is found. The maximum number of retries for the `CustomRetryPlugin` is increased from 6 to 10.
Co-authored-by: Hangfei Lin <hangfei@google.com>
PiperOrigin-RevId: 830984481
Merge https://github.com/google/adk-python/pull/2167, during which:
- Let the already added `max_query_result_rows` field cover for `max_downloaded_rows` field added in the PR
- Revert `max_rows` parameter added to execute_sql function, the tool config control should serve most practical use cases
- Keep the relevant tests for tool config
🤖 Generated with [Claude Code](https://claude.ai/code)
COPYBARA_INTEGRATE_REVIEW=https://github.com/google/adk-python/pull/2167 from lupuletic:feature/configurable-max-downloaded-rows 23c56905c297d7aec2be4f1eb86ea23c8178bf21
PiperOrigin-RevId: 830701093
Enhanced `save_artifact` in `callback_context.py` to accept `custom_metadata` and added `get_artifact_version` to retrieve artifact details.
Introduced a new sample, `context_offloading_with_artifact`, demonstrating how to use ADK artifacts to offload large data from the LLM context. The sample includes:
- `QueryLargeDataTool`: Generates mock sales reports, saves them as artifacts with custom metadata, and injects the artifact content into the LLM request immediately after creation.
- `CustomLoadArtifactsTool`: Provides summaries of available artifacts to the LLM based on metadata and loads artifact content on demand when `load_artifacts` is called.
Co-authored-by: Hangfei Lin <hangfei@google.com>
PiperOrigin-RevId: 830592786
Merge https://github.com/google/adk-python/pull/2889
# Implement Full async DatabaseSessionService
**Target Issue:** #1005
## Overview
This PR introduces an asynchronous implementation of the `DatabaseSessionService` with minimal breaking changes. The primary goal is to enable effective use of ADK in fully async environments and API endpoints while avoiding event loop blocking during database I/O operations.
## Changes
- Converted `DatabaseSessionService` to use async/await patterns throughout
## Testing Plan
The implementation has been tested following the project's contribution guidelines:
### Unit Tests
- All existing unit tests pass successfully
- Minor update to test requirements added to support `aiosqlite`
### Manual End-to-End Testing
- E2E tests performed using:
- **LLM Provider:** LiteLLM
- **Database:** PostgreSQL with `asyncpg` driver
```python
from google.adk.sessions.database_session_service import DatabaseSessionService
connection_string: str = (
"postgresql+asyncpg://PG_USER:PG_PSWD@PG_HOST:5432/PG_DB"
)
session_service: DatabaseSessionService = DatabaseSessionService(
db_url=connection_string
)
session = await session_service.create_session(
app_name="test_app", session_id="test_session", user_id="test_user"
)
assert session is not None
sessions = await session_service.list_sessions(app_name="test_app", user_id="test_user")
assert len(sessions.sessions) > 0
session = await session_service.get_session(
app_name="test_app", session_id="test_session", user_id="test_user"
)
assert session is not None
await session_service.delete_session(
app_name="test_app", session_id="test_session", user_id="test_user"
)
assert (
await session_service.get_session(
app_name="test_app", session_id="test_session", user_id="test_user"
)
is None
)
```
The implementation have been also tested using the following configurations for llm provider and Runner:
```python
def get_azure_openai_model(deployment_id: str | None = None) -> LiteLlm:
...
if not deployment_id:
deployment_id = os.getenv("AZURE_OPENAI_DEPLOYMENT_ID")
logger.info(f"Using Azure OpenAI deployment ID: {deployment_id}")
return LiteLlm(
model=f"azure/{os.getenv('AZURE_OPENAI_DEPLOYMENT_ID')}",
stream=True,
)
...
@staticmethod
def _get_runner(agent: Agent) -> Runner:
storage=DatabaseSessionService(db_url=get_pg_connection_string())
return Runner(
agent=agent,
app_name=APP_NAME,
session_service=storage,
)
...
async for event in self.runner.run_async(
user_id=user_id,
session_id=session_id,
new_message=content,
run_config=(
RunConfig(
streaming_mode=StreamingMode.SSE, response_modalities=["TEXT"]
)
if stream
else RunConfig()
),
):
last_event = event
if stream:
yield event
...
```
## Breaking Changes
- Database connection string format may need updates for async drivers
Co-authored-by: Shangjie Chen <deanchen@google.com>
COPYBARA_INTEGRATE_REVIEW=https://github.com/google/adk-python/pull/2889 from GitMarco27:feature/async_database_session_service e1b1b14934c1fb7975a6832cdd1549e94acab985
PiperOrigin-RevId: 830525148
The tool description now clarifies that `load_artifacts` should be called when a user uploads files, to make those artifacts accessible within the session.
Co-authored-by: George Weale <gweale@google.com>
PiperOrigin-RevId: 830469802
The upper bound for the `fastapi` dependency is adjusted from `<1.119.0` to `<0.119.0`.
Close#3422
Co-authored-by: George Weale <gweale@google.com>
PiperOrigin-RevId: 830465681
Update `_function_declaration_to_tool_param` to use `getattr` when accessing the `required` field within `function_declaration.parameters`. This prevents `AttributeError` when a `FunctionDeclaration`'s parameters schema does not include a `required` attribute
Close#2891
Co-authored-by: George Weale <gweale@google.com>
PiperOrigin-RevId: 830225582
Merge https://github.com/google/adk-python/pull/3352
Replace send() method with send_realtime_input() to fix DeprecationWarning
### Link to Issue or Description of Change
**1. Link to an existing issue (if applicable):**
- Closes: #2393
### Testing Plan
Have run unit test for test_live_request_queue.py
**Unit Tests:**
- [ ] I have added or updated unit tests for my change.
- [x] All unit tests pass locally.
Summary :
tests/unittests/agents/test_live_request_queue.py::test_send_realtime PASSED [100%]
**Manual End-to-End (E2E) Tests:**
### Checklist
- [x] I have read the [CONTRIBUTING.md](https://github.com/google/adk-python/blob/main/CONTRIBUTING.md) document.
- [x] I have performed a self-review of my own code.
- [x] I have commented my code, particularly in hard-to-understand areas.
- [ ] I have added tests that prove my fix is effective or that my feature works.
- [x] New and existing unit tests pass locally with my changes.
- [x] I have manually tested my changes end-to-end.
- [ ] Any dependent changes have been merged and published in downstream modules.
Tested both API variants.
Co-authored-by: Hangfei Lin <hangfei@google.com>
COPYBARA_INTEGRATE_REVIEW=https://github.com/google/adk-python/pull/3352 from SanjaySiddharth:fix-dep-session.send b23b641a63ff20d1f2dbd9abd519f8facf0aab77
PiperOrigin-RevId: 830182844
Merge https://github.com/google/adk-python/pull/2875
related: https://github.com/google/adk-python/issues/2876
## Reason for this change:
Currently, there is no direct way to access the user_id from within agent contexts, plugins, or callbacks. This limitation prevents several important use cases:
1. **User-specific logging and tracing**: When debugging or monitoring agent behavior, it's crucial to associate actions with specific users for better observability.
2. **User-scoped operations**: Plugins and callbacks often need to perform user-specific operations, such as accessing user-specific resources or applying user-level configurations.
3. **Session management**: The user_id is a key component of session identification, but it's not accessible through the ReadonlyContext interface, requiring workarounds to access it.
## Changes made:
- Added a `user_id` property to the `ReadonlyContext` class in `src/google/adk/agents/readonly_context.py`
- The property exposes the user_id from the underlying invocation context as a readonly field
## Impact:
This change will:
- Enable plugins and callbacks to access the current user's ID directly through the context
- Improve logging and tracing capabilities by allowing user-specific tracking
- Simplify code that needs to perform user-scoped operations without requiring access to internal implementation details
- Maintain backward compatibility as this is an additive change
### Before:
- No direct way to access user_id from ReadonlyContext
### After:
```python
@property
def user_id(self) -> str:
"""The id of the user. READONLY field."""
return self._invocation_context.user_id
```
This is a non-breaking change that adds a new readonly property to the existing ReadonlyContext interface.
Co-authored-by: Hangfei Lin <hangfei@google.com>
COPYBARA_INTEGRATE_REVIEW=https://github.com/google/adk-python/pull/2875 from ammmr:ammmr-add-user_id-property-to-readonly-context 771734ebf39cb0748a5dc826436236268fbb58f8
PiperOrigin-RevId: 830170908
The new Sqlite version has fixed schema and use a single column to store Event data, this should avoid DB migration for future add to the Event object.
- This change introduces `SqliteSessionService`, an asynchronous session service using `aiosqlite` that stores event data as JSON within SQLite.
- A migration script, `migrate_from_sqlalchemy_sqlite.py`, is included to transition data from the older SQLAlchemy-based SQLite schema to this new format.
- The CLI service registry is updated to use SqliteSessionService for sqlite:// URIs.
- Throw error when user trying to access a legacy DB and advice the user to do the migration.
Co-authored-by: Shangjie Chen <deanchen@google.com>
PiperOrigin-RevId: 829971174
NOTE:
- This is the expected behavior, so that SSE streaming and non-streaming behaviors are idential to each other in the persisted session.
- The underlying google_llm.py hasn't been fully comply with this behavior, which will be addressed in future changes.
Co-authored-by: Wei Sun (Jack) <weisun@google.com>
PiperOrigin-RevId: 829878175
This change updates the `BigQueryAgentAnalyticsPlugin` (formerly in `bigquery_logging_plugin.py`) to perform BigQuery writes asynchronously in background tasks, preventing blocking of the main agent execution flow. Initialization of BigQuery clients and table creation is also made asynchronous. The content logged in various callbacks has been streamlined and simplified. The plugin file has been renamed to `bigquery_agent_analytics_plugin.py` to match the class name.
PiperOrigin-RevId: 829603260
Merge https://github.com/google/adk-python/pull/2600
**Reason for this change:**
Currently, the detailed llm_request object is logged at the info level when connecting to the live API. This causes two main issues:
1. Excessive log output with large instructions: When instructions are large, the unstructured log output becomes problematic. Since the logs are not structured, any
newline characters in the instruction content cause each line to be output as a separate log entry, resulting in massive log output that floods the logging system.
2. Unnecessary verbosity in production: The detailed request information is primarily useful for debugging purposes and should not appear in standard info-level logs,
especially when connections are established frequently.
**Changes made:**
- Changed logger.info to logger.debug in src/google/adk/models/google_llm.py at line 294 for the live connection logging statement.
**Impact:**
This change will:
- Reduce log noise in production environments where info-level logging is enabled
- Keep the detailed connection information available when debug-level logging is needed for troubleshooting
- Improve log readability by showing only essential information at the info level
**Before:**
logger.info('Connecting to live with llm_request:%s', llm_request)
**After:**
logger.debug('Connecting to live with llm_request:%s', llm_request)
This is a non-breaking change that only affects logging verbosity.
Co-authored-by: Hangfei Lin <hangfei@google.com>
COPYBARA_INTEGRATE_REVIEW=https://github.com/google/adk-python/pull/2600 from ammmr:ammmr-info-to-debug-connect-log 36eab157f5d5ca00b44ee6ad8a4e1aeec871600c
PiperOrigin-RevId: 829596092
We need to set project and location parameters when we instantiate the Google GenAI library Client for VertexAi. This used to work before through environment variables but it doesn't seem to work anymore. So we are updating the `ApigeeLlm` wrapper to read from `GOOGLE_CLOUD_PROJECT` and `GOOGLE_CLOUD_LOCATION` environment variables and pass to the Client constructor.
PiperOrigin-RevId: 829569362
This trimmed schema includes only the fields relevant to agent shells, tool wiring, and common generation parameters, improving efficiency and focus.
The default model for the assistant has also been updated to "gemini-2.5-pro"
Co-authored-by: George Weale <gweale@google.com>
PiperOrigin-RevId: 829513627
Merge https://github.com/google/adk-python/pull/3127
# What this PR does
This PR fixes a bug in the `execute_sql function` within `bigquery/query_tool.py`. The bug caused an error when executing queries that do not return a result set, such as a `CREATE TEMP FUNCTION` statement.
# The problem
Within `execute_sql,` when a user executed a query that doesn't produce a result set, like the following:
```sql
CREATE TEMP FUNCTION f() AS (0); SELECT f();
```
The corresponding BigQuery job object would have its `destination` property set to `None`. This caused an error because the code was written with the expectation of a valid destination table.
# The solution
The fix modifies `execute_sql` to check if the `query_job.destination` is None. If it is, the function correctly identifies the query as a statement that doesn't require a destination and allows it to complete successfully.
This ensures that `execute_sql` can robustly handle these types of queries.
COPYBARA_INTEGRATE_REVIEW=https://github.com/google/adk-python/pull/3127 from na0fu3y:patch-1 69218a8d25699dfa2e18bb305087f01192e91e91
PiperOrigin-RevId: 829270885
This change adds a debug log statement to output the function response IDs and the event list when a matching function call event is not found, which helps in debugging.
Co-authored-by: Hangfei Lin <hangfei@google.com>
PiperOrigin-RevId: 829059186
Capture the function call ID from events and use it to populate the `function_response.id` in subsequent user messages, enabling recording and replaying of test cases involving long-running tools.
Co-authored-by: Liang Wu <wuliang@google.com>
PiperOrigin-RevId: 829042356
Add `_to_litellm_response_format` to convert ADK's `response_schema` types (Pydantic models, JSON schema dicts) into the format needed by LiteLLM for JSON object/schema constraints
Close#1967
Co-authored-by: George Weale <gweale@google.com>
PiperOrigin-RevId: 829037987
In `gemini_llm_connection.py`, accumulate partial transcription texts and emit `LlmResponse` with `partial=True` for each chunk. When the transcription is marked as `finished`, emit a final `LlmResponse` with the full accumulated text and `partial=False`.
In `runners.py`, modify `_should_append_to_history` to only add transcription events to the history when they are fully finished, preventing partial transcriptions from being added.
Co-authored-by: Hangfei Lin <hangfei@google.com>
PiperOrigin-RevId: 829029715
This is about 35% decrease. This change refactors several ADK modules to import `vertexai` and its submodules only when they are first used, rather than at the top of the file. This improves module load times by avoiding unnecessary imports of large dependencies. Imports are also placed within `if TYPE_CHECKING:` blocks where appropriate.
Co-authored-by: Liang Wu <wuliang@google.com>
PiperOrigin-RevId: 829017293
This accounts for about 8% of latency in cold start, because DatabaseSessionService imports from sqlalchemy. After the change, DatabaseSessionService is only imported if it's really needed by the program. The other objects in this file are not affected because they contribute to a small fraction of the latency.
Co-authored-by: Liang Wu <wuliang@google.com>
PiperOrigin-RevId: 829009868
Merge https://github.com/google/adk-python/pull/3324
**Problem:**
ADK seems to not pass output/input transcription `finished` flag from the Gemini.
**Solution:**
Relaxation of checking conditions of valid llm_response for input/output transcription.
### Testing Plan
Unit test `test_receive_transcript_finished` checks if input/output transcription message with no text but `finished` flag is received.
**Unit Tests:**
- [x] I have added or updated unit tests for my change.
- [x] All unit tests pass locally.
<img width="785" height="373" alt="image" src="https://github.com/user-attachments/assets/6d870e9f-1372-4808-91a9-38578c1b3729" />
**Manual End-to-End (E2E) Tests:**
Configure Gemini Agent to produce input & output transcriptions. Observe incoming transcript messages - to see if the finished flag appears at the end agents & users statement.
### Checklist
- [x] I have read the [CONTRIBUTING.md](https://github.com/google/adk-python/blob/main/CONTRIBUTING.md) document.
- [x] I have performed a self-review of my own code.
- [x] I have commented my code, particularly in hard-to-understand areas.
- [x] I have added tests that prove my fix is effective or that my feature works.
- [x] New and existing unit tests pass locally with my changes.
- [x] I have manually tested my changes end-to-end.
- [x] Any dependent changes have been merged and published in downstream modules.
### Additional context
The mentioned `finished` flag can be obtained in native Gemini APIs like via Websocket but not via ADK.
Co-authored-by: Hangfei Lin <hangfei@google.com>
COPYBARA_INTEGRATE_REVIEW=https://github.com/google/adk-python/pull/3324 from ChrisQlasty:fix/transcript_finish e7b8e5e5f5fac1d2dfd495f2dbadb19ed4328b0c
PiperOrigin-RevId: 828987996
SaveFilesAsArtifactsPlugin now resolves the canonical URI for each saved artifact. When the URI is model-accessible (`gs://`, `https://`, `http://`), we add a `Part(file_data=...)` so the LLM can fetch the filedirectly while still emitting the placeholder. If no model-accessible URI exists, we retain the original inline blob alongside the placeholder to preserve access to the uploaded bytes.
Close#2016
Co-authored-by: George Weale <gweale@google.com>
PiperOrigin-RevId: 828786519
The `_run_compaction_default` method is no longer called and can be safely removed.
Co-authored-by: Hangfei Lin <hangfei@google.com>
PiperOrigin-RevId: 828760157
Merge https://github.com/google/adk-python/pull/3403
This PR corrects misspellings identified by the [check-spelling action](https://github.com/marketplace/actions/check-spelling)
Note: while I use tooling to identify errors, the tooling doesn't _actually_ provide the corrections, I'm picking them on my own. I'm a human, and I may make mistakes.
### Testing Plan
The misspellings have been reported at https://github.com/jsoref/adk-python/actions/runs/19056081305/attempts/1#summary-54426435973
The action reports that the changes in this PR would make it happy: https://github.com/jsoref/adk-python/actions/runs/19056081446/attempts/1#summary-54426436321
**Unit Tests:**
- [ ] I have added or updated unit tests for my change.
- [ ] All unit tests pass locally.
_Please include a summary of passed `pytest` results._
**Manual End-to-End (E2E) Tests:**
_Please provide instructions on how to manually test your changes, including any
necessary setup or configuration. Please provide logs or screenshots to help
reviewers better understand the fix._
### Checklist
- [x] I have read the [CONTRIBUTING.md](https://github.com/google/adk-python/blob/main/CONTRIBUTING.md) document.
- [x] I have performed a self-review of my own code.
- [ ] I have commented my code, particularly in hard-to-understand areas.
- [ ] I have added tests that prove my fix is effective or that my feature works.
- [ ] New and existing unit tests pass locally with my changes.
- [ ] I have manually tested my changes end-to-end.
- [ ] Any dependent changes have been merged and published in downstream modules.
### Additional context
- https://github.com/google/adk-python/pull/3382#issuecomment-3488654110
COPYBARA_INTEGRATE_REVIEW=https://github.com/google/adk-python/pull/3403 from jsoref:spelling-markdown 432b7ded3bddfe4a1d8127c50fd7a001f19bda83
PiperOrigin-RevId: 828754459
Merge https://github.com/google/adk-python/pull/2582
Relate to #3306
## Description
Fixes issues #6 and #1785 where ADK commands crash on Windows due to symlink creation requiring admin privileges.
## Problem
On Windows, running ADK commands (like `adk run`) fails with `OSError: [WinError 1314] A required privilege is not held by the client: ...` because the logging system attempts to create a symlink for the latest log file. Windows requires administrator privileges for symlink creation by default (unless Developer Mode is enabled), causing crashes for non-admin users.
## Root Cause
The issue was in [`logs.py`](https://github.com/google/adk-python/blob/main/src/google/adk/cli/utils/logs.py#L72) where `os.symlink()` was called unconditionally without error handling, causing the entire CLI to crash when symlink creation failed.
## Solution
This PR implements graceful symlink handling. Changes made:
- Extracted symlink creation into separate function with error handling
- Added graceful fallback when symlink creation fails
- Replaced crashes with warnings to keep CLI functional
- Improved user messaging about log file locations
This solution follows a similar pattern to the one used in [ROS2](https://github.com/ros2) in its [logging module](https://github.com/ros2/launch/blob/rolling/launch/launch/logging/__init__.py#L93).
## Testing
### Before (broken)
```bash
> adk run my_agent
Log setup complete: C:\Users\username\AppData\Local\Temp\agents_log\agent.20250817_215119.log
Traceback (most recent call last):
File "google\adk\cli\utils\logs.py", line 72, in log_to_tmp_folder
os.symlink(log_filepath, latest_log_link)
OSError: [WinError 1314] A required privilege is not held by the client
```
### After (fixed)
```bash
> adk run my_agent
Log setup complete: C:\Users\username\AppData\Local\Temp\agents_log\agent.20250817_215319.log
UserWarning: Cannot create symlink for latest log file: [WinError 1314] A required privilege is not held by the client
To access latest log: tail -F C:\Users\username\AppData\Local\Temp\agents_log\agent.20250817_215319.log
Running agent my_agent, type exit to exit.
```
Co-authored-by: Wei Sun (Jack) <weisun@google.com>
COPYBARA_INTEGRATE_REVIEW=https://github.com/google/adk-python/pull/2582 from lorenzofavaro:fix/windows-symlink-permissions 1e99a6287994f1bfd9ac11550d3dc06205998df5
PiperOrigin-RevId: 828690483
Merge https://github.com/google/adk-python/pull/3382
This PR corrects misspellings identified by the [check-spelling action](https://github.com/marketplace/actions/check-spelling)
Note: while I use tooling to identify errors, the tooling doesn't _actually_ provide the corrections, I'm picking them on my own. I'm a human, and I may make mistakes.
### Testing Plan
The misspellings have been reported at https://github.com/jsoref/adk-python/actions/runs/19056081305/attempts/1#summary-54426435973
The action reports that the changes in this PR would make it happy: https://github.com/jsoref/adk-python/actions/runs/19056081446/attempts/1#summary-54426436321
**Unit Tests:**
- [ ] I have added or updated unit tests for my change.
- [ ] All unit tests pass locally.
_Please include a summary of passed `pytest` results._
**Manual End-to-End (E2E) Tests:**
_Please provide instructions on how to manually test your changes, including any
necessary setup or configuration. Please provide logs or screenshots to help
reviewers better understand the fix._
### Checklist
- [x] I have read the [CONTRIBUTING.md](https://github.com/google/adk-python/blob/main/CONTRIBUTING.md) document.
- [x] I have performed a self-review of my own code.
- [ ] I have commented my code, particularly in hard-to-understand areas.
- [ ] I have added tests that prove my fix is effective or that my feature works.
- [ ] New and existing unit tests pass locally with my changes.
- [ ] I have manually tested my changes end-to-end.
- [ ] Any dependent changes have been merged and published in downstream modules.
### Additional context
- follow-up to #2447
Co-authored-by: Hangfei Lin <hangfei@google.com>
COPYBARA_INTEGRATE_REVIEW=https://github.com/google/adk-python/pull/3382 from jsoref:spelling 3df5932e97391c437e1638d7163e5e3d5c52d5c3
PiperOrigin-RevId: 828686941
Merge https://github.com/google/adk-python/pull/3394
This PR corrects misspellings identified by the [check-spelling action](https://github.com/marketplace/actions/check-spelling)
Note: while I use tooling to identify errors, the tooling doesn't _actually_ provide the corrections, I'm picking them on my own. I'm a human, and I may make mistakes.
### Testing Plan
The misspellings have been reported at https://github.com/jsoref/adk-python/actions/runs/19056081305/attempts/1#summary-54426435973
The action reports that the changes in this PR would make it happy: https://github.com/jsoref/adk-python/actions/runs/19056081446/attempts/1#summary-54426436321
**Unit Tests:**
- [ ] I have added or updated unit tests for my change.
- [ ] All unit tests pass locally.
_Please include a summary of passed `pytest` results._
**Manual End-to-End (E2E) Tests:**
_Please provide instructions on how to manually test your changes, including any
necessary setup or configuration. Please provide logs or screenshots to help
reviewers better understand the fix._
### Checklist
- [x] I have read the [CONTRIBUTING.md](https://github.com/google/adk-python/blob/main/CONTRIBUTING.md) document.
- [x] I have performed a self-review of my own code.
- [ ] I have commented my code, particularly in hard-to-understand areas.
- [ ] I have added tests that prove my fix is effective or that my feature works.
- [ ] New and existing unit tests pass locally with my changes.
- [ ] I have manually tested my changes end-to-end.
- [ ] Any dependent changes have been merged and published in downstream modules.
### Additional context
- https://github.com/google/adk-python/pull/3382#issuecomment-3488654110
COPYBARA_INTEGRATE_REVIEW=https://github.com/google/adk-python/pull/3394 from jsoref:spelling-contributing c3d5e342c4350f7cae9f8f0c6638b176f2e30e80
PiperOrigin-RevId: 828659867
This commit refactors the `BigQueryAgentAnalyticsPlugin` to leverage the modern BigQuery Storage Write API, replacing the previous implementation that used the legacy `insert_rows_json` method (based on `tabledata.insertAll`).
**Key Changes:**
* **Switched to Storage Write API:** Event logs are now ingested using the `BigQueryWriteClient` from the `google-cloud-bigquery-storage` library.
* **Utilizes Default Stream:** We are using the `_default` stream for sending data, which is an efficient method for streaming in data without needing to manage stream lifecycles. This is ideal for continuous event logging.
* **Apache Arrow Format:** Log entries are converted to the Apache Arrow format using `pyarrow` before being sent. The BigQuery table schema is dynamically converted to an Arrow schema. This binary format is more efficient than JSON.
* **Updated Initialization:** The plugin now initializes both the standard `bigquery.Client` (for table management) and the `BigQueryWriteClient`.
* **Test Updates:** Unit tests in `test_bigquery_logging_plugin.py` have been comprehensively updated to mock the new `BigQueryWriteClient`, `bq_schema_utils`, and `pyarrow` components. Tests now verify calls to `append_rows` and the data structure passed to create the Arrow RecordBatch.
**Benefits of this change:**
* **Improved Performance:** The Storage Write API is designed for high-throughput streaming and offers better performance compared to the legacy API.
* **Reduced Cost:** Ingesting data via the Storage Write API is generally more cost-effective.
* **Enhanced Reliability:** The Storage Write API provides more robust streaming capabilities.
* **Modernization:** Aligns the plugin with the recommended best practices for BigQuery data ingestion.
This change enhances the efficiency and scalability of the BigQuery logging plugin.
PiperOrigin-RevId: 828655496
Merge https://github.com/google/adk-python/pull/3411
## Summary
Fixes AgentTool cleanup to prevent MCP session errors by calling `runner.close()` after sub-agent execution.
## Problem
When using AgentTool with MCP tools, the runner cleanup happened during garbage collection in a different async task context, causing:
```
RuntimeError: Attempted to exit cancel scope in a different task than it was entered in
```
## Solution
- Call `await runner.close()` immediately after sub-agent execution in AgentTool
- This ensures MCP sessions and other resources are cleaned up in the correct async task context
- Updated test mock to include the close() method
## Demo Agents
Added two comprehensive demo agents showing how to use AgentTool with MCP tools:
### mcp_in_agent_tool_remote (SSE mode)
- Uses HTTP/SSE connection to remote MCP server
- Zero-installation setup with `uvx`
- Demonstrates server-side MCP deployment pattern
### mcp_in_agent_tool_stdio (stdio mode)
- Uses subprocess connection with automatic server launch
- Fully automatic setup with `uvx` subdirectory syntax
- Demonstrates embedded MCP deployment pattern
Both demos:
- Use Gemini 2.5 Flash
- Include example prompts for JSON Schema exploration
- Have comprehensive READMEs with architecture diagrams
- Follow ADK agent structure conventions
## Testing
- âś… All existing unit tests pass
- âś… Manual testing with both SSE and stdio modes
- âś… Verified cleanup happens in correct async context
- âś… No more cancel scope errors with MCP tools
## Related
- Fixes#1112
- Related to #929
Co-authored-by: Wei Sun (Jack) <weisun@google.com>
COPYBARA_INTEGRATE_REVIEW=https://github.com/google/adk-python/pull/3411 from google:fix/agent-tool-mcp-cleanup 9ae753b5a4
PiperOrigin-RevId: 828651896
Currently ADK web hangs and logs "AGSI callable returned without completing response" when the server is unreachable. To fix, set timeouts for connecting to server.
Co-authored-by: Kathy Wu <wukathy@google.com>
PiperOrigin-RevId: 828648928
- add FileArtifactService that persists artifacts to the local filesystem
- adjust BaseArtifactService and exports so callers can wire in the filebacked implementation
Co-authored-by: George Weale <gweale@google.com>
PiperOrigin-RevId: 828629298
Replace the full JSON schema dump with a compact text summary of key AgentConfig components like LlmAgent, ToolConfig, and GenerateContentConfig
Co-authored-by: George Weale <gweale@google.com>
PiperOrigin-RevId: 828627911
Merge https://github.com/google/adk-python/pull/3407
This PR corrects misspellings identified by the [check-spelling action](https://github.com/marketplace/actions/check-spelling)
Note: while I use tooling to identify errors, the tooling doesn't _actually_ provide the corrections, I'm picking them on my own. I'm a human, and I may make mistakes.
### Testing Plan
The misspellings have been reported at https://github.com/jsoref/adk-python/actions/runs/19056081305/attempts/1#summary-54426435973
The action reports that the changes in this PR would make it happy: https://github.com/jsoref/adk-python/actions/runs/19056081446/attempts/1#summary-54426436321
**Unit Tests:**
- [ ] I have added or updated unit tests for my change.
- [ ] All unit tests pass locally.
_Please include a summary of passed `pytest` results._
**Manual End-to-End (E2E) Tests:**
_Please provide instructions on how to manually test your changes, including any
necessary setup or configuration. Please provide logs or screenshots to help
reviewers better understand the fix._
### Checklist
- [x] I have read the [CONTRIBUTING.md](https://github.com/google/adk-python/blob/main/CONTRIBUTING.md) document.
- [x] I have performed a self-review of my own code.
- [ ] I have commented my code, particularly in hard-to-understand areas.
- [ ] I have added tests that prove my fix is effective or that my feature works.
- [ ] New and existing unit tests pass locally with my changes.
- [ ] I have manually tested my changes end-to-end.
- [ ] Any dependent changes have been merged and published in downstream modules.
### Additional context
- https://github.com/google/adk-python/pull/3382#issuecomment-3488654110
COPYBARA_INTEGRATE_REVIEW=https://github.com/google/adk-python/pull/3407 from jsoref:spelling-issue-template ce8febc6ab5368640c0ec2ae5d6a2c8ac5bf8ed6
PiperOrigin-RevId: 828610865
* Fix issue with MCP tools throwing an error ([1a4261a](https://github.com/google/adk-python/commit/1a4261ad4b66cdeb39d39110a086bd6112b17516))
* Remove redundant `format` field from LiteLLM content objects ([489c39d](https://github.com/google/adk-python/commit/489c39db01465e38ecbc2c7f32781c349b8cddc9))
* Update the contribution analysis tool to use original write mode ([54db3d4](https://github.com/google/adk-python/commit/54db3d4434e0706b83a589fa2499d11d439a6e4e))
* Update dependency version constraints to be based on PyPI versions([0b1784e](https://github.com/google/adk-python/commit/0b1784e0e493a0e2df1edfe37e5ed5f4247e7d9d))
### Improvements
@@ -97,13 +99,14 @@
* Add sample agent for VertexAiCodeExecutor ([edfe553](https://github.com/google/adk-python/commit/edfe5539421d196ca4da14d3a37fac7b598f8c8d))
* Adds a new sample agent that demonstrates how to integrate PostgreSQL databases using the Model Context Protocol (MCP) ([45a2168](https://github.com/google/adk-python/commit/45a2168e0e6773e595ecfb825d7e4ab0a38c3a38))
* Add example for using ADK with Fast MCP sampling ([d3796f9](https://github.com/google/adk-python/commit/d3796f9b33251d28d05e6701f11e80f02a2a49e1))
* Refactor gepa sample code and clean-up user demo colab([63353b2](https://github.com/google/adk-python/commit/63353b2b74e23e97385892415c5a3f2a59c3504f))
* Add a service registry to provide a generic way to register custom service implementations to be used in FastAPI server. See short instruction [here](https://github.com/google/adk-python/discussions/3175#discussioncomment-14745120). ([391628f](https://github.com/google/adk-python/commit/391628fcdc7b950c6835f64ae3ccab197163c990))
* Add a service registry to provide a generic way to register custom service implementations to be used in FastAPI server. See [short instruction](https://github.com/google/adk-python/discussions/3175#discussioncomment-14745120). ([391628f](https://github.com/google/adk-python/commit/391628fcdc7b950c6835f64ae3ccab197163c990))
* Add the ability to rewind a session to before a previous invocation ([9dce06f](https://github.com/google/adk-python/commit/9dce06f9b00259ec42241df4f6638955e783a9d1))
* Support resuming a parallel agent with multiple branches paused on tool confirmation requests ([9939e0b](https://github.com/google/adk-python/commit/9939e0b087094038b90d86c2fd35c26dd63f1157))
* Support content union as static instruction ([cc24d61](https://github.com/google/adk-python/commit/cc24d616f80c0eba2b09239b621cf3d176f144ea))
@@ -170,9 +173,8 @@
* Set default for `bypass_multi_tools_limit` to False for GoogleSearchTool and VertexAiSearchTool ([6da7274](https://github.com/google/adk-python/commit/6da727485898137948d72906d86d78b6db6331ac))
* Add more clear instruction to the doc updater agent about one PR for each recommended change ([b21d0a5](https://github.com/google/adk-python/commit/b21d0a50d610407be2f10b73a91274840ffdfe18))
* Add a guideline to avoid content deletion ([16b030b](https://github.com/google/adk-python/commit/16b030b2b25a9b0b489e47b4b148fc4d39aeffcb))
* Add an sample agent for the `ReflectAndRetryToolPlugin` ([9b8a4aa](https://github.com/google/adk-python/commit/9b8a4aad6fe65ef37885e5c3368d2799a2666534))
* Add a sample agent for the `ReflectAndRetryToolPlugin` ([9b8a4aa](https://github.com/google/adk-python/commit/9b8a4aad6fe65ef37885e5c3368d2799a2666534))
* Improve error message when adk web is run in wrong directory ([4a842c5](https://github.com/google/adk-python/commit/4a842c5a1334c3ee01406f796651299589fe12ab))
* Add an sample agent for the `ReflectAndRetryToolPlugin` ([9b8a4aa](https://github.com/google/adk-python/commit/9b8a4aad6fe65ef37885e5c3368d2799a2666534))
* Add span for context caching handling and new cache creation ([a2d9f13](https://github.com/google/adk-python/commit/a2d9f13fa1d31e00ba9493fba321ca151cdd9366))
* Disable the scheduled execution for issue triage workflow ([bae2102](https://github.com/google/adk-python/commit/bae21027d9bd7f811bed638ecce692262cb33fe5))
* Correct the callback signatures ([fa84bcb](https://github.com/google/adk-python/commit/fa84bcb5756773eadff486b99c9bd416b4faa9c6))
@@ -202,7 +204,7 @@
* Add `dry_run` functionality to BigQuery `execute_sql` tool ([960eda3](https://github.com/google/adk-python/commit/960eda3d1f2f46dc93a365eb3de03dc3483fe9bb))
* Support Oauth2 client credentials grant type ([5c6cdcd](https://github.com/google/adk-python/commit/5c6cdcd197a6780fc86d9183fa208f78c8a975d9))
* Support OAuth2 client credentials grant type ([5c6cdcd](https://github.com/google/adk-python/commit/5c6cdcd197a6780fc86d9183fa208f78c8a975d9))
* Add `ReflectRetryToolPlugin` to reflect from errors and retry with different arguments when tool errors ([e55b894](https://github.com/google/adk-python/commit/e55b8946d6a2e01aaf018d6a79d11d13c5286152))
* Support using `VertexAiSearchTool` built-in tool with other tools in the same agent ([4485379](https://github.com/google/adk-python/commit/4485379a049a5c84583a43c85d444ea1f1ba6f12))
* Support using google search built-in tool with other tools in the same agent ([d3148da](https://github.com/google/adk-python/commit/d3148dacc97f0a9a39b6d7a9640f7b7b0d6f9a6c))
@@ -283,7 +285,7 @@
* Add endpoint to generate memory from session ([2595824](https://github.com/google/adk-python/commit/25958242db890b4d2aac8612f7f7cfbb561727fa))
* **[Tools]**
* Add Google Maps Grounding Tool to ADK ([6b49391](https://github.com/google/adk-python/commit/6b493915469ecb42068e24818ab547b0856e4709))
* **MCP:** Initialize tool_name_prefix in MCPToolse ([86dea5b](https://github.com/google/adk-python/commit/86dea5b53ac305367283b7e353b60d0f4515be3b))
* **MCP:** Initialize tool_name_prefix in MCPToolset ([86dea5b](https://github.com/google/adk-python/commit/86dea5b53ac305367283b7e353b60d0f4515be3b))
* **[Evals]**
* Data model for storing App Details and data model for steps ([01923a9](https://github.com/google/adk-python/commit/01923a9227895906ca8ae32712d65b178e2cd7d5))
* Adds Rubric based final response evaluator ([5a485b0](https://github.com/google/adk-python/commit/5a485b01cd64cb49735e13ebd5e7fa3da02cd85f))
@@ -350,7 +352,7 @@
* Fix discussion answering github action workflow to escape the quote in the discussion content JSON [43c9681](https://github.com/google/adk-python/commit/43c96811da891a5b0c9cf1be525665e65f346a13)
* Send full MIME types for image/video/pdf in get_content [e45c3be](https://github.com/google/adk-python/commit/e45c3be23895b5ec68908ad9ee19bd622dcbd003)
* Fix flaky unit tests: tests/unittests/flows/llm_flows/test_functions_simple.py [b92b288](https://github.com/google/adk-python/commit/b92b288c978a9b3d1a76c8bcb96cc8f439ce610b)
* Make UT of a2a consistent about how tests should be skipped when python verison < 3.10 [98b0426](https://github.com/google/adk-python/commit/98b0426cd2dc5e28014ead22b22dbf50d42d0a9a)
* Make UT of a2a consistent about how tests should be skipped when python version < 3.10 [98b0426](https://github.com/google/adk-python/commit/98b0426cd2dc5e28014ead22b22dbf50d42d0a9a)
### Improvements
@@ -441,7 +443,7 @@ with Bigtable for building AI Agent applications(experimental feature) ([a953807
* Using base event's invocation id when merge multiple function response event ([279e4fe](https://github.com/google/adk-python/commit/279e4fedd0b1c0d1499c0f9a4454357af7da490e))
* Avoid crash when there is no candidates_token_count, which is Optional ([22f34e9](https://github.com/google/adk-python/commit/22f34e9d2c552fbcfa15a672ef6ff0c36fa32619))
* Fix the packaging version comparison logic in adk cli ([a2b7909](https://github.com/google/adk-python/commit/a2b7909fc36e7786a721f28e2bf75a1e86ad230d))
* Fixes SequentialAgent.config_type type hint ([8a9a271](https://github.com/google/adk-python/commit/8a9a271141678996c9b84b8c55d4b539d011391c))
* Fixes the host in the ansi bracket of adk web ([cd357bf](https://github.com/google/adk-python/commit/cd357bf5aeb01f1a6ae2a72349a73700ca9f1ed2))
* Add spanner tool name prefix ([a27927d](https://github.com/google/adk-python/commit/a27927dc8197c391c80acb8b2c23d610fba2f887))
@@ -453,7 +455,7 @@ with Bigtable for building AI Agent applications(experimental feature) ([a953807
* Update `openai` dependency version, based on correct OPENAI release ([bb8ebd1](https://github.com/google/adk-python/commit/bb8ebd15f90768b518cd0e21a59b269e30d6d944))
* Add the missing license header for core_callback_config init file ([f8fd6a4](https://github.com/google/adk-python/commit/f8fd6a4f09ab520b8ecdbd8f9fe48228dbff7ebe))
* Creates yaml_utils.py in utils to allow adk dump yaml in the same style ([1fd58cb](https://github.com/google/adk-python/commit/1fd58cb3633992cd88fa7e09ca6eda0f9b34236f))
* Return explict None type for DELETE endpoints ([f03f167](https://github.com/google/adk-python/commit/f03f1677790c0a9e59b6ba6f46010d0b7b64be50))
* Return explicit None type for DELETE endpoints ([f03f167](https://github.com/google/adk-python/commit/f03f1677790c0a9e59b6ba6f46010d0b7b64be50))
* Add _config suffix to all yaml-based agent examples ([43f302c](https://github.com/google/adk-python/commit/43f302ce1ab53077ee8f1486d5294540678921e6))
* Rename run related method and request to align with the conventions ([ecaa7b4](https://github.com/google/adk-python/commit/ecaa7b4c9847b478c7cdc37185b1525f733bb403))
* Update models in samples/ folder to be gemini 2.0+ ([6c217ba](https://github.com/google/adk-python/commit/6c217bad828edf62b41ec06b168f8a6cb7ece2ed))
@@ -478,7 +480,7 @@ with Bigtable for building AI Agent applications(experimental feature) ([a953807
### Bug Fixes
* A2A RPC URL got overriden by host and port param of adk api server ([52284b1](https://github.com/google/adk-python/commit/52284b1bae561e0d6c93c9d3240a09f210551b97))
* A2A RPC URL got overridden by host and port param of adk api server ([52284b1](https://github.com/google/adk-python/commit/52284b1bae561e0d6c93c9d3240a09f210551b97))
* Aclose all async generators to fix OTel tracing context ([a30c63c](https://github.com/google/adk-python/commit/a30c63c5933a770b960b08a6e2f8bf13eece8a22))
* Use PreciseTimestamp for create and update time in database session service to improve precision ([585141e](https://github.com/google/adk-python/commit/585141e0b7dda20abb024c7164073862c8eea7ae))
* Ignore AsyncGenerator return types in function declarations ([e2518dc](https://github.com/google/adk-python/commit/e2518dc371fe77d7b30328d8d6f5f864176edeac))
@@ -506,7 +508,7 @@ with Bigtable for building AI Agent applications(experimental feature) ([a953807
* Group FastAPI endpoints with tags ([c323de5](https://github.com/google/adk-python/commit/c323de5c692223e55372c3797e62d4752835774d))
* Allow implementations to skip defining a close method on Toolset ([944e39e](https://github.com/google/adk-python/commit/944e39ec2a7c9ad7f20c08fd66bf544de94a23d7))
* Add sample agent to test support of output_schema and tools at the same time for gemini model ([f2005a2](https://github.com/google/adk-python/commit/f2005a20267e1ee8581cb79c37aa55dc8e18c0ea))
* Add Github workflow config for uploading ADK docs to knowledge store ([5900273](https://github.com/google/adk-python/commit/59002734559d49a46940db9822b9c5f490220a8c))
* Add GitHub workflow config for uploading ADK docs to knowledge store ([5900273](https://github.com/google/adk-python/commit/59002734559d49a46940db9822b9c5f490220a8c))
* Update ADK Answering agent to reference doc site instead of adk-docs repo ([b5a8bad](https://github.com/google/adk-python/commit/b5a8bad170e271b475385dac440c7983ed207df8))
@@ -32,7 +33,7 @@ deployment-agnostic, and compatible with other frameworks.
## 🔥 What's new
- **Custom Service Registration**: Add a service registry to provide a generic way to register custom service implementations to be used in FastAPI server. See short instruction [here](https://github.com/google/adk-python/discussions/3175#discussioncomment-14745120). ([391628f](https://github.com/google/adk-python/commit/391628fcdc7b950c6835f64ae3ccab197163c990))
- **Custom Service Registration**: Add a service registry to provide a generic way to register custom service implementations to be used in FastAPI server. See [short instruction](https://github.com/google/adk-python/discussions/3175#discussioncomment-14745120). ([391628f](https://github.com/google/adk-python/commit/391628fcdc7b950c6835f64ae3ccab197163c990))
- **Rewind**: Add the ability to rewind a session to before a previous invocation ([9dce06f](https://github.com/google/adk-python/commit/9dce06f9b00259ec42241df4f6638955e783a9d1)).
You are a helper bot that updates ADK docs in Github Repository {DOC_OWNER}/{DOC_REPO}
based on the code in the ADK Python codebase in Github Repository {CODE_OWNER}/{CODE_REPO} according to the instructions in the ADK docs issues.
You are a helper bot that updates ADK docs in GitHub Repository {DOC_OWNER}/{DOC_REPO}
based on the code in the ADK Python codebase in GitHub Repository {CODE_OWNER}/{CODE_REPO} according to the instructions in the ADK docs issues.
You are very familiar with Github, expecially how to search for files in a Github repository using git grep.
You are very familiar with GitHub, especially how to search for files in a GitHub repository using git grep.
# 2. Responsibilities
Your core responsibility includes:
@@ -69,18 +69,18 @@ root_agent = Agent(
# 3. Workflow
1. Always call the `clone_or_pull_repo` tool to make sure the ADK docs and codebase repos exist in the local folder {LOCAL_REPOS_DIR_PATH}/repo_name and are the latest version.
2. Read and analyze the issue specified by user.
- If user only specified the issue number, call the `get_issue` tool to get the issue details, otherwise use the issue details provided by user directly.
- If user only specified the issue number, call the `get_issue` tool to get the issue details; otherwise, use the issue details provided by user directly.
3. If the issue contains instructions about how to update the ADK docs, follow the instructions to update the ADK docs.
4. Understand the doc update instructions.
- Ignore and skip the instructions about updating API reference docs, since it will be automatically generated by the ADK team.
5. Read the doc to update using the `read_local_git_repo_file_content` tool from the local ADK docs repo under {LOCAL_REPOS_DIR_PATH}/{DOC_REPO}.
6. Find the related Python files in the ADK Python codebase.
- If the doc update instructions specify paths to the Python files, use them directly, otherwise use a list of regex search patterns to find the related Python files through the `search_local_git_repo` tool.
- If the doc update instructions specify paths to the Python files, use them directly; otherwise, use a list of regex search patterns to find the related Python files through the `search_local_git_repo` tool.
- You should focus on the main ADK Python codebase, ignore the changes in tests or other auxiliary files.
- You should find all the related Python files, not only the most relevant one.
7. Read the specified or found Python files using the `read_local_git_repo_file_content` tool to find all the related code.
- You can ignore unit test files, unless you are sure that the test code is uesful to understand the related concepts.
- You should read all the the found files to find all the related code, unless you already know the content of the file or you are sure that the file is not related to the ADK doc.
- You can ignore unit test files, unless you are sure that the test code is useful to understand the related concepts.
- You should read all the found files to find all the related code, unless you already know the content of the file or you are sure that the file is not related to the ADK doc.
8. Update the ADK doc file according to the doc update instructions and the related code.
- Use active voice phrasing in your doc updates.
- Use second person "you" form of address in your doc updates.
@@ -102,12 +102,12 @@ root_agent = Agent(
- **File Paths:** Always use absolute paths when calling the tools to read files, list directories, or search the codebase.
- **Tool Call Parallelism:** Execute multiple independent tool calls in parallel when feasible (i.e. searching the codebase).
- **Avoid deletion:** Do not delete any existing content unless specifically directed to do so.
- **Explaination:** Provide concise explanations for your actions and reasoning for each step.
- **Explanation:** Provide concise explanations for your actions and reasoning for each step.
- **Minimize changes:** When making updates to documentation pages, make the minimum amount of changes to achieve the communication goal. Only make changes that are necessary, and leave everything else as-is.
- **Avoid trivial code sample changes:** Update code samples only when adding or modifying functionality. Do not reformat code samples, change variable names, or change code syntax unless you are specifically directed to make those updates.
# 5. Output
Present the followings in an easy to read format as the final output to the user.
Present the following in an easy to read format as the final output to the user.
You are a helper bot that checks if ADK docs in Github Repository {DOC_REPO} owned by {DOC_OWNER}
should be updated based on the changes in the ADK Python codebase in Github Repository {CODE_REPO} owned by {CODE_OWNER}.
You are a helper bot that checks if ADK docs in GitHub Repository {DOC_REPO} owned by {DOC_OWNER}
should be updated based on the changes in the ADK Python codebase in GitHub Repository {CODE_REPO} owned by {CODE_OWNER}.
You are very familiar with Github, expecially how to search for files in a Github repository using git grep.
You are very familiar with GitHub, especially how to search for files in a GitHub repository using git grep.
# 2. Responsibilities
Your core responsibility includes:
- Find all the code changes between the two ADK releases.
- Find **all** the related docs files in ADK Docs repository under the "/docs/" directory.
- Compare the code changes with the docs files and analyze the differences.
- Write the instructions about how to update the ADK docs in markdown format and create a Github issue in the Github Repository {DOC_REPO} with the instructions.
- Write the instructions about how to update the ADK docs in markdown format and create a GitHub issue in the GitHub Repository {DOC_REPO} with the instructions.
# 3. Workflow
1. Always call the `clone_or_pull_repo` tool to make sure the ADK docs and codebase repos exist in the local folder {LOCAL_REPOS_DIR_PATH}/repo_name and are the latest version.
@@ -102,8 +102,8 @@ root_agent = Agent(
**Reference**:
Reference to the code file (e.g. src/google/adk/tools/spanner/metadata_tool.py).
```
- When referncing doc file, use the full relative path of the doc file in the ADK Docs repository (e.g. docs/sessions/memory.md).
9. Create or recommend to create a Github issue in the Github Repository {DOC_REPO} with the instructions using the `create_issue` tool.
- When referencing doc file, use the full relative path of the doc file in the ADK Docs repository (e.g. docs/sessions/memory.md).
9. Create or recommend to create a GitHub issue in the GitHub Repository {DOC_REPO} with the instructions using the `create_issue` tool.
- The title of the issue should be "Found docs updates needed from ADK python release <start_tag> to <end_tag>", where start_tag and end_tag are the release tags.
- The body of the issue should be the instructions about how to update the ADK docs.
- Include the compare link between the two ADK releases in the issue body, e.g. https://github.com/google/adk-python/compare/v1.14.0...v1.14.1.
@@ -112,7 +112,7 @@ root_agent = Agent(
# 4. Guidelines & Rules
- **File Paths:** Always use absolute paths when calling the tools to read files, list directories, or search the codebase.
- **Tool Call Parallelism:** Execute multiple independent tool calls in parallel when feasible (i.e. searching the codebase).
- **Explaination:** Provide concise explanations for your actions and reasoning for each step.
- **Explanation:** Provide concise explanations for your actions and reasoning for each step.
- **Reference:** For each recommended change, reference the code changes (i.e. links to the commits) **AND** the code files (i.e. relative paths to the code files in the codebase).
- **Sorting:** Sort the recommended changes by the importance of the changes, from the most important to the least important.
- Here are the importance groups: Feature changes > Bug fixes > Other changes.
@@ -122,7 +122,7 @@ root_agent = Agent(
- If a change in the codebase can be covered by the auto-generated API reference docs, you should just recommend to update the API reference docs (i.e. regenerate the API reference docs) instead of the other human-written ADK docs.
# 5. Output
Present the followings in an easy to read format as the final output to the user.
Present the following in an easy to read format as the final output to the user.
sandbox, `agent_engine_id`). If the issue does not explicitly mention
Agent Engine concepts, do not use this label—choose "core" instead.
- "a2a": Agent-to-agent workflows, coordination logic, or A2A protocol.
- "bq": BigQuery integration or general issues related to BigQuery.
When unsure between labels, prefer the most specific match. If a label
cannot be assigned confidently, do not call the labeling tool.
@@ -214,7 +216,8 @@ root_agent = Agent(
- Use "agent engine" only when the issue clearly references Vertex AI Agent Engine deployment artifacts (for example `.agent_engine_config.json`, `ae_ignore`, `agent_engine_id`, or Agent Engine sandbox errors).
- If it's about Model Context Protocol (e.g. MCP tool, MCP toolset, MCP session management etc.), label it with both "mcp" and "tools".
- If it's about A2A integrations or workflows, label it with "a2a".
- If you can't find a appropriate labels for the issue, follow the previous instruction that starts with "IMPORTANT:".
- If it's about BigQuery integrations, label it with "bq".
- If you can't find an appropriate labels for the issue, follow the previous instruction that starts with "IMPORTANT:".
Call the `add_label_and_owner_to_issue` tool to label the issue, which will also assign the issue to the owner of the label.
@@ -231,7 +234,7 @@ root_agent = Agent(
- Mention the assigned owner when a label maps to one.
- If no label is applied, clearly state why.
Present the followings in an easy to read format highlighting issue number and your label.
Present the following in an easy to read format highlighting issue number and your label.
- the issue summary in a few sentence
- your label recommendation and justification
- the owner of the label if you assign the issue to an owner
@@ -9,7 +9,7 @@ This sample data science agent uses Agent Engine Code Execution Sandbox to execu
* 1. Follow https://cloud.google.com/vertex-ai/generative-ai/docs/agent-engine/code-execution/overview to create a code execution sandbox environment.
* 2. Replace the SANDBOX_RESOURCE_NAME with the one you just created. If you dont want to create a new sandbox environment directly, the Agent Engine Code Execution Sandbox will create one for you by default using the AGENT_ENGINE_RESOURCE_NAME you specified, however, please ensure to clean up sandboxes after use, otherwise, it will consume quotas.
* 2. Replace the SANDBOX_RESOURCE_NAME with the one you just created. If you dont want to create a new sandbox environment directly, the Agent Engine Code Execution Sandbox will create one for you by default using the AGENT_ENGINE_RESOURCE_NAME you specified, however, please ensure to clean up sandboxes after use; otherwise, it will consume quotas.
**Statefulness:** All code snippets are executed and the variables stays in the environment. You NEVER need to re-initialize variables. You NEVER need to reload files. You NEVER need to re-import libraries.
**Output Visibility:** Always print the output of code execution to visualize results, especially for data exploration and analysis. For example:
You can read about the Auth Code grant / flow type in detail [here](https://developer.okta.com/blog/2018/04/10/oauth-authorization-code-grant-type). But for the purpose of this demo, following steps take place
You can read about the [Auth Code grant / flow type](https://developer.okta.com/blog/2018/04/10/oauth-authorization-code-grant-type) in detail. But for the purpose of this demo, following steps take place
1. The user asks the agent to find hotels in "New York".
2. Agent realizes (based on LLM response) that it needs to call a tool and that the tool needs authentication.
# The environment variables OAUTH_CLIENT_ID and OAUTH_CLIENT_SECRET
# must be set
credentials_config=BigtableCredentialsConfig(
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Google Team Member
George Weale
Ankur Sharma
Yifan Wang
Shangjie Chen
Hangfei Lin
Mimi Sun
Max Ind
Michael Jones
Kathy Wu
Xuan Yang
Wei Sun (Jack)
anrzeszutek
milkisbad
Yeesian Ng
Sara Robinson
Liang Wu
Catalin Lupuleti
GitMarco27
Amy Wu
Jeff Bryner
Sanjay Siddharth MJ
GenkiNoguchi
Lê Nam Khánh
Henrique Cadioli
Naofumi Yamada
Josh Soref
Krzysztof Czuszynski
Kapil Sachdeva
ananyablonko
Lorenzo
Wei Sun (Jack)