SEP-1686: Tasks (#1645) · modelcontextprotocol/python-sdk@c92bb2f

File tree

76 files changed

lines changed

- - simple-task-interactive-client
    - mcp_simple_task_interactive_client
- - - mcp_simple_task_interactive

76 files changed

lines changed

Original file line number	Diff line number	Diff line change
`@@ -2512,6 +2512,7 @@ MCP servers declare capabilities during initialization:`
`2512`	`2512`	`## Documentation`
`2513`	`2513`
`2514`	`2514`	`- [API Reference](https://modelcontextprotocol.github.io/python-sdk/api/)`
	`2515`	`+- [Experimental Features (Tasks)](https://modelcontextprotocol.github.io/python-sdk/experimental/tasks/)`
`2515`	`2516`	`- [Model Context Protocol documentation](https://modelcontextprotocol.io)`
`2516`	`2517`	`- [Model Context Protocol specification](https://modelcontextprotocol.io/specification/latest)`
`2517`	`2518`	`- [Officially supported servers](https://github.com/modelcontextprotocol/servers)`

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,43 @@`
	`1`	`+# Experimental Features`
	`2`	`+`
	`3`	`+!!! warning "Experimental APIs"`
	`4`	`+`
	`5`	`+The features in this section are experimental and may change without notice.`
	`6`	`+They track the evolving MCP specification and are not yet stable.`
	`7`	`+`
	`8`	`+This section documents experimental features in the MCP Python SDK. These features`
	`9`	`+implement draft specifications that are still being refined.`
	`10`	`+`
	`11`	`+## Available Experimental Features`
	`12`	`+`
	`13`	`+### [Tasks](tasks.md)`
	`14`	`+`
	`15`	`+Tasks enable asynchronous execution of MCP operations. Instead of waiting for a`
	`16`	`+long-running operation to complete, the server returns a task reference immediately.`
	`17`	`+Clients can then poll for status updates and retrieve results when ready.`
	`18`	`+`
	`19`	`+Tasks are useful for:`
	`20`	`+`
	`21`	`+- Long-running computations that would otherwise block`
	`22`	`+- Batch operations that process many items`
	`23`	`+- Interactive workflows that require user input (elicitation) or LLM assistance (sampling)`
	`24`	`+`
	`25`	`+## Using Experimental APIs`
	`26`	`+`
	`27`	+Experimental features are accessed via the `.experimental` property:
	`28`	`+`
	`29`	+```python
	`30`	`+# Server-side`
	`31`	`+@server.experimental.get_task()`
	`32`	`+async def handle_get_task(request: GetTaskRequest) -> GetTaskResult:`
	`33`	`+...`
	`34`	`+`
	`35`	`+# Client-side`
	`36`	`+result = await session.experimental.call_tool_as_task("tool_name", {"arg": "value"})`
	`37`	+```
	`38`	`+`
	`39`	`+## Providing Feedback`
	`40`	`+`
	`41`	`+Since these features are experimental, feedback is especially valuable. If you encounter`
	`42`	`+issues or have suggestions, please open an issue on the`
	`43`	`+[python-sdk repository](https://github.com/modelcontextprotocol/python-sdk/issues).`