|
1 | 1 | # Changelog
|
2 | 2 |
|
| 3 | +## 1.1.0 |
| 4 | + |
| 5 | +### Features |
| 6 | + |
| 7 | +* Extract agent loading logic from fast_api.py to a separate AgentLoader class and support more agent definition folder/file structure. |
| 8 | +* Added audio play in web UI. |
| 9 | +* Added input transcription support for live/streaming. |
| 10 | +* Added support for storing eval run history locally in adk eval cli. |
| 11 | +* Image artifacts can now be clicked directly in chat message to view. |
| 12 | +* Left side panel can now be resized. |
| 13 | + |
| 14 | +### Bug Fixes |
| 15 | + |
| 16 | +* Avoid duplicating log in stderr. |
| 17 | +* Align event filtering and ordering logic. |
| 18 | +* Add handling for None param.annotation. |
| 19 | +* Fixed several minor bugs regarding eval tab in web UI. |
| 20 | + |
| 21 | +### Miscellaneous Chores |
| 22 | + |
| 23 | +* Updates mypy config in pyproject.toml. |
| 24 | +* Add google search agent in samples. |
| 25 | +* Update filtered sch
10000
ema parameters for Gemini API. |
| 26 | +* Adds autoformat.sh for formatting codebase. |
| 27 | + |
| 28 | +## 1.0.0 |
| 29 | + |
| 30 | +### ⚠ BREAKING CHANGES |
| 31 | + |
| 32 | +* Evaluation dataset schema is finalized with strong-type pydantic models. |
| 33 | + (previously saved eval file needs re-generation, for both adk eval cli and |
| 34 | + the eval tab in adk web UI). |
| 35 | +* `BuiltInCodeExecutor` (in code_executors package) replaces |
| 36 | + `BuiltInCodeExecutionTool` (previously in tools package). |
| 37 | +* All methods in services are now async, including session service, artifact |
| 38 | + service and memory service. |
| 39 | + * `list_events` and `close_session` methods are removed from session service. |
| 40 | +* agent.py file structure with MCP tools are now easier and simpler ([now](https://github.com/google/adk-python/blob/3b5232c14f48e1d5b170f3698d91639b079722c8/contributing/samples/mcp_stdio_server_agent/agent.py#L33) vs [before](https://github.com/google/adk-python/blob/a4adb739c0d86b9ae4587547d2653d568f6567f2/contributing/samples/mcp_agent/agent.py#L41)). |
| 41 | + Old format is not working anymore. |
| 42 | +* `Memory` schema and `MemoryService` is redesigned. |
| 43 | +* Mark various class attributes as private in the classes in the `tools` package. |
| 44 | +* Disabled session state injection if instruction provider is used. |
| 45 | + (so that you can have `{var_name}` in the instruction, which is required for code snippets) |
| 46 | +* Toolbox integration is revamped: tools/toolbox_tool.py → tools/toolbox_toolset.py. |
| 47 | +* Removes the experimental `remote_agent.py`. We'll redesign it and bring it back. |
| 48 | + |
| 49 | +### Features |
| 50 | + |
| 51 | +* Dev UI: |
| 52 | + * A brand new trace view for overall agent invocation. |
| 53 | + * A revamped evaluation tab and comparison view for checking eval results. |
| 54 | +* Introduced `BaseToolset` to allow dynamically add/remove tools for agents. |
| 55 | + * Revamped MCPToolset with the new BaseToolset interface. |
| 56 | + * Revamped GoogleApiTool, GoogleApiToolset and ApplicationIntegrationToolset with the new BaseToolset interface. |
| 57 | + * Resigned agent.py file structure when needing MCPToolset. |
| 58 | + * Added ToolboxToolset. |
| 59 | +* Redesigned strong-typed agent evaluation schema. |
| 60 | + * Allows users to create more cohesive eval sets. |
| 61 | + * Allows evals to be extended for non-text modality. |
| 62 | + * Allows for a structured interaction with the uber eval system. |
| 63 | +* Redesigned Memory schema and MemoryService interfaces. |
| 64 | +* Added token usage to LlmResponse. |
| 65 | +* Allowed specifying `--adk_version` in `adk deploy cloud_run` cli. Default is the current version. |
| 66 | + |
| 67 | +### Bug Fixes |
| 68 | + |
| 69 | +* Fixed `adk deploy cloud_run` failing bug. |
| 70 | +* Fixed logs not being printed due to `google-auth` library. |
| 71 | + |
| 72 | +### Miscellaneous Chores |
| 73 | + |
| 74 | +* Display full help text when adk cli receives invalid arguments. |
| 75 | +* `adk web` now binds `127.0.0.1` by default, instead of 0.0.0.0. |
| 76 | +* `InMemoryRunner` now takes `BaseAgent` in constructor. |
| 77 | +* Various docstring improvements. |
| 78 | +* Various UI tweaks. |
| 79 | +* Various bug fixes. |
| 80 | +* Update various contributing/samples for contributors to validate the implementation. |
| 81 | + |
| 82 | + |
| 83 | +## 0.5.0 |
| 84 | + |
| 85 | +### ⚠ BREAKING CHANGES |
| 86 | + |
| 87 | +* Updated artifact and memory service interface to be async. Agents that |
| 88 | + interact with these services through callbacks or tools will now need to |
| 89 | + adjust their invocation methods to be async (using await), or ensure calls |
| 90 | + are wrapped in an asynchronous executor like asyncio.run(). Any service that |
| 91 | + extends the base interface must also be updated. |
| 92 | + |
| 93 | +### Features |
| 94 | + |
| 95 | +* Introduced the ability to chain model callbacks. |
| 96 | +* Added support for async agent and model callbacks. |
| 97 | +* Added input transcription support for live/streaming. |
| 98 | +* Captured all agent code error and display on UI. |
| 99 | +* Set param required tag to False by default in openapi_tool. |
| 100 | +* Updated evaluation functions to be asynchronous. |
| 101 | + |
| 102 | +### Bug Fixes |
| 103 | + |
| 104 | +* Ensured a unique ID is generated for every event. |
| 105 | +* Fixed the issue when openapi_specparser has parameter.required as None. |
| 106 | +* Updated the 'type' value on the items/properties nested structures for Anthropic models to adhere to JSON schema. |
| 107 | +* Fix litellm error issues. |
| 108 | + |
| 109 | +### Miscellaneous Chores |
| 110 | + |
| 111 | +* Regenerated API docs. |
| 112 | +* Created a `developer` folder and added samples. |
| 113 | +* Updated the contributing guide. |
| 114 | +* Docstring improvements, typo fixings, GitHub action to enforce code styles on formatting and imports, etc. |
| 115 | + |
| 116 | +## 0.4.0 |
| 117 | + |
| 118 | +### ⚠ BREAKING CHANGES |
| 119 | +* Set the max size of strings in database columns. MySQL mandates that all VARCHAR-type fields must specify their lengths. |
| 120 | +* Extract content encode/decode logic to a shared util, resolve issues with JSON serialization, and update key length for DB table to avoid key too long issue in mysql. |
| 121 | +* Enhance `FunctionTool` to verify if the model is providing all the mandatory arguments. |
| 122 | + |
| 123 | +### Features |
| 124 | +* Update ADK setup guide to improve onboarding experience. |
| 125 | +* feat: add ordering to recent events in database session service. |
| 126 | +* feat(llm_flows): support async before/after tool callbacks. |
| 127 | +* feat: Added --replay and --resume options to adk run cli. Check adk run --help for more details. |
| 128 | +* Created a new Integration Connector Tool (underlying of the ApplicationIntegrationToolSet) so that we do not force LLM to provide default value. |
| 129 | + |
| 130 | +### Bug Fixes |
| 131 | + |
| 132 | +* Don't send content with empty text to LLM. |
| 133 | +* Fix google search reading undefined for `renderedContent`. |
| 134 | + |
| 135 | +### Miscellaneous Chores |
| 136 | +* Docstring improvements, typo fixings, github action to enfore code styles on formatting and imports, etc. |
| 137 | + |
3 | 138 | ## 0.3.0
|
4 | 139 |
|
5 | 140 | ### ⚠ BREAKING CHANGES
|
6 | 141 |
|
7 | 142 | * Auth: expose `access_token` and `refresh_token` at top level of auth
|
8 |
| - credentails, instead of a `dict` |
| 143 | + credentials, instead of a `dict` |
9 | 144 | ([commit](https://github.com/google/adk-python/commit/956fb912e8851b139668b1ccb8db10fd252a6990)).
|
10 | 145 |
|
11 | 146 | ### Features
|
|
28 | 163 |
|
29 | 164 | ### Miscellaneous Chores
|
30 | 165 |
|
31 |
| -* README.md impprovements. |
| 166 | +* README.md improvements. |
32 | 167 | * Various code improvements.
|
33 | 168 | * Various typo fixes.
|
34 | 169 | * Bump min version of google-genai to 1.11.0.
|
|
86 | 221 | * Built-in evaluation support
|
87 | 222 | * Development UI that makes local development easy
|
88 | 223 | * Deploy to Google Cloud Run, Agent Engine
|
89 |
| -* (Experimental) Live(Bidi) auido/video agent support and Compositional Function Calling(CFC) support |
| 224 | +* (Experimental) Live(Bidi) audio/video agent support and Compositional Function Calling(CFC) support |
0 commit comments