* feat: Update Safety and Security page to include plugins as a method for implement agent-agnostic security
* Add hyperlinks to newly added plugin section
---------
Co-authored-by: sammalik-lgtm <sammalik@google.com>
* refactor: swap to StdioConnectionParams
- Fix import path to avoid ImportError
- Changed from StdioServerParameters to StdioConnectionParams as recommended.
* fix: Re-modify imports for MCP config libraries
---------
Co-authored-by: Joe Fernandez <joefernandez@users.noreply.github.com>
the removed line is not needed -events is never used and the 2 lines below is creating events_async
Co-authored-by: Joe Fernandez <joefernandez@users.noreply.github.com>
* feat: Update Safety and Security page to include plugins as a method for implement agent-agnostic security
* Add hyperlinks to newly added plugin section
---------
Co-authored-by: Joe Fernandez <joefernandez@users.noreply.github.com>
The CICD tests responsible for generating static pages are failing to add the necessary keys to the site. The current settings proved unhelpful, so they have been cleared.
* fix: update broken authenticated tool helper
* fix: Update invalid auth config argument reference in helpers
* fix: Incorrect code example for `get_auth_config`
`auth_request_function_call`'s args values are dictionaries, meaning it will never be an instance of AuthConfig by default
This commit clarifies the behavior of the `temp:` namespace in the documentation.
The following changes were made:
- Updated `docs/sessions/state.md` to provide a more detailed explanation of the `temp:` namespace.
- Updated `docs/context/index.md` to improve the clarity of an example related to passing data between tools.
- Updated `docs/runtime/index.md` to add a note about the scope of `temp:` variables.
- Updated `docs/agents/multi-agents.md` and `docs/agents/workflow-agents/sequential-agents.md` to clarify how the `temp:` namespace is propagated to sub-agents.
- Updated `docs/tools/function-tools.md` to explain how to use the `temp:` namespace to pass data between tool calls.
fixes#160
Co-authored-by: google-labs-jules[bot] <161369871+google-labs-jules[bot]@users.noreply.github.com>
Co-authored-by: Joe Fernandez <joefernandez@users.noreply.github.com>
This commit adds a new page to the documentation for the REST API, with
details about the `/run` and `/run_sse` endpoints.
A new file is created at `docs/api-reference/rest/index.md` and linked
from the main API reference page and the navigation bar.
Joe: Modified navigation ordering, and added information to point to REST API guide in Testing topic.
fixes#126
Co-authored-by: google-labs-jules[bot] <161369871+google-labs-jules[bot]@users.noreply.github.com>
Co-authored-by: Joe Fernandez <joefernandez@users.noreply.github.com>
* Add comprehensive MCP deployment documentation
- Add detailed deployment section for ADK agents with MCP tools
- Document critical synchronous agent definition requirement for deployment
- Cover three deployment patterns: self-contained stdio, remote SSE/HTTP, and sidecar containers
- Include production deployment checklist with security, monitoring, and scalability considerations
- Provide environment-specific configurations for Cloud Run, GKE, and Vertex AI Agent Engine
- Add troubleshooting guide for common MCP deployment issues
- Include quick deployment commands for Agent Engine and Cloud Run
Addresses deployment challenges outlined in issue #327 where async MCP toolset creation
is no longer supported in deployment environments.
Fixes#327
* Replace SSE/HTTP with Streamable HTTP implementation in MCP Pattern 2
- Update Pattern 2 deployment example to use StreamableHTTPSessionManager
- Replace basic SSE server code with comprehensive Streamable HTTP implementation
- Add production-ready features: stateless mode, ASGI compliance, lifecycle management
- Include Starlette application setup with proper error handling
The `lookup_order_status` example in the tool documentation was using the `status` field to represent both the tool's execution outcome and the business-level order status. This is ambiguous and can be confusing for the LLM.
This change refactors the example to use a dedicated `status` field for the execution outcome (`success`/`error`) and nests the order-specific data in a separate `order` dictionary. This aligns with best practices for tool design and improves clarity.
Fixes#530
Co-authored-by: google-labs-jules[bot] <161369871+google-labs-jules[bot]@users.noreply.github.com>
Co-authored-by: Joe Fernandez <joefernandez@users.noreply.github.com>
In the human-in-the-loop example, the long_running_function_response
variable could be overwritten with `None` by a subsequent event that
did not contain a function response. This would cause the logic that
depends on the response to fail.
Fixes#375
Co-authored-by: google-labs-jules[bot] <161369871+google-labs-jules[bot]@users.noreply.github.com>
Co-authored-by: Joe Fernandez <joefernandez@users.noreply.github.com>
Yeesian Ng