DaprMCPWorkflowClient and MCP Workflow Helpers
Add DaprMCPWorkflowClient to the Python SDK so all Python AI frameworks (dapr-agents, python-ai, etc.) can use it without depending on dapr-agents.
Scope
DaprMCPWorkflowClient class: connect(server_name), get_all_tools(), call_tool()
- Starts ListTools/CallTool workflows via
DaprClient.start_workflow
- Returns tools as typed objects with input schemas
- Workflow name construction (
dapr.internal.mcp.<server>.<method>)
- JSON-compatible types matching proto definitions
- Reuse
create_pydantic_model_from_schema for tool arg models
- Unit tests: mock
DaprClient.start_workflow, assert correct workflow names/inputs
- Example usage and docs
Blocked by
Blocks
- dapr-agents MCPServer integration (consumes this client)
- diagridio/python-ai MCP support
Related
DaprMCPWorkflowClient and MCP Workflow Helpers
Add
DaprMCPWorkflowClientto the Python SDK so all Python AI frameworks (dapr-agents, python-ai, etc.) can use it without depending on dapr-agents.Scope
DaprMCPWorkflowClientclass:connect(server_name),get_all_tools(),call_tool()DaprClient.start_workflowdapr.internal.mcp.<server>.<method>)create_pydantic_model_from_schemafor tool arg modelsDaprClient.start_workflow, assert correct workflow names/inputsBlocked by
Blocks
Related