github.com-google-adk-python

test: add unit tests for public symbols that had no coverage Co-authored-by: George Weale <gweale@google.com> PiperOrigin-RevId: 960421043

refactor(types): make the agents, sessions, flows and workflow packages pass strict mypy Not annotations-only. This is one component's slice of a repo-wide typing cleanup, and the wider change was found to contain behavior changes that have not all been individually triaged, so please review it as a functional change. Co-authored-by: George Weale <gweale@google.com> PiperOrigin-RevId: 958554313

refactor(types): make google.adk.models pass strict mypy Not annotations-only. This is one component's slice of a repo-wide typing cleanup, and the wider change was found to contain behavior changes that have not all been individually triaged, so please review it as a functional change. Co-authored-by: George Weale <gweale@google.com> PiperOrigin-RevId: 958623646

refactor(types): type the integrations, skills and MCP tool packages for strict mypy Co-authored-by: George Weale <gweale@google.com> PiperOrigin-RevId: 961125207

feat: support context caching 1. add a context cache config in app level which will apply to all agents in the app 2. pass on cache config through invocation context to llm_reqeust 3. store cache metadata in llm_response 4. lookup old cache metadata from latest event for reusing old cache 5. create new cache if old cache cannot be reused PiperOrigin-RevId: 809158578

feat: Add ComputerUseToolset PiperOrigin-RevId: 785646071

refactor(types): make google.adk.a2a pass strict mypy Not annotations-only. This is one component's slice of a repo-wide typing cleanup, and the wider change was found to contain behavior changes that have not all been individually triaged, so please review it as a functional change. Co-authored-by: George Weale <gweale@google.com> PiperOrigin-RevId: 959932563

feat: add OpenAI Responses API support in labs Add OpenAIResponsesLlm and AzureOpenAIResponsesLlm under labs.openai, a BaseLlm targeting the OpenAI Responses API: request/response/streaming conversion, reasoning summaries, structured output, tool mapping, and usage metadata. Merge https://github.com/google/adk-python/pull/6188 Closes #3209 Co-authored-by: Luca Frigato <37444661+FrigaZzz@users.noreply.github.com> Co-authored-by: George Weale <gweale@google.com> PiperOrigin-RevId: 938272066

feat: Add Vertex AI Agent Engine Sandbox integration for computer use Implement AgentEngineSandboxComputer, a BaseComputer implementation that uses Vertex AI Agent Engine Computer Use Sandbox as a remote browser environment. This enables computer-use agents to operate in secure, isolated cloud-based browser environments. Key features: - SandboxClient: Low-level CDP client using vertexai SDK send_command() - AgentEngineSandboxComputer: BaseComputer implementation with session state management for sandbox sharing across invocations - Support for both auto-provisioning and bring-your-own-sandbox (BYOS) modes - Automatic token refresh and retry logic for transient navigation errors - Fix wait_5_seconds adapter to properly pass tool_context Co-authored-by: Xiang (Sean) Zhou <seanzhougoogle@google.com> PiperOrigin-RevId: 900292803

Create a developer folder and add samples. PiperOrigin-RevId: 755885332

chore: Add initial version of agent builder assistant that assists user to build config based agent PiperOrigin-RevId: 801014241

feat(telemetry): add request-driven metric export for Agent Engine On the Vertex AI Agent Runtime CPU is throttled the instant a request finishes, so a background periodic metric exporter is starved between requests and drops data. This adds a request-driven metric reader (and the span processor + middleware that drive it) that collects and exports metrics on the request path, where CPU is guaranteed, without adding latency to any request. The default GCP metric exporter is also switched to a raw OTLP push exporter over telemetry.googleapis.com. Co-authored-by: Max Ind <maxind@google.com> PiperOrigin-RevId: 955921308

feat: Support interactions API for calling models Co-authored-by: Xiang (Sean) Zhou <seanzhougoogle@google.com> PiperOrigin-RevId: 843032402

feat: Adds Rubric based final response evaluator The evaluator uses a set of rubrics to assess the quality of the agent's final response. PiperOrigin-RevId: 811154498

fix: Support project-based gemini model path for BuiltInCodeExecutor and all built-in tools This is to support model path like : projects/265104255505/locations/us-central1/publishers/google/models/gemini-2.0-flash-001" PiperOrigin-RevId: 783413351

fix: Secure and harden FileArtifactService against tampered metadata and partial writes Co-authored-by: Xuan Yang <xygoogle@google.com> PiperOrigin-RevId: 960747887

feat(mcp): gracefully handle tool execution errors and transport crashes Previously, if an MCP tool returned a JSON-RPC error (e.g. 403 Forbidden) or if the underlying transport connection crashed, the resulting exceptions (McpError and ConnectionError) would bubble up and crash the entire ADK runner. This change introduces robust error boundaries for MCP tools: - `McpTool.run_async()` now catches `McpError` and general exceptions, returning them as structured error dictionaries `{"error": ...}` to the LLM agent so the conversation can continue gracefully. - `SessionContext` races tool calls against the background session task so transport crashes surface immediately instead of hanging. - Fixes an AnyIO cancellation scope bug ("Attempted to exit cancel scope in a different task") by removing redundant `asyncio.wait_for` wrappers around exit stack context entry. - Connection errors trigger automatic retries via `@retry_on_errors` before finally surfacing the failure to the agent. Fixes #4901, #4162 Co-authored-by: Kathy Wu <wukathy@google.com> PiperOrigin-RevId: 902912292

feat: Add agent card builder PiperOrigin-RevId: 785565665

refactor(types): make the code executors, planners and runner pass strict mypy Not annotations-only. This is one component's slice of a repo-wide typing cleanup, and the wider change was found to contain behavior changes that have not all been individually triaged, so please review it as a functional change. Co-authored-by: George Weale <gweale@google.com> PiperOrigin-RevId: 958458433

fix(telemetry): mark failed tool spans as errors and fix provider naming Co-authored-by: George Weale <gweale@google.com> PiperOrigin-RevId: 957285853