Fix pyupgrade and AGENTS.md reconnect description

- pyupgrade: drop forward-reference string annotations in _mcp.py (Python 3.10+ resolves them natively now that MCPTaskOptions is defined before use).

- AGENTS.md: align reconnect description with current behavior. Phase 1 (initial tools/call) does NOT retry on connection loss; raises 'connection lost; task state unknown' instead, so a server that accepted the request but lost the response cannot start the operation twice. Phase 2 (tasks/get / tasks/result) still reconnects once against the same task_id.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

python/packages/core/AGENTS.md

@@ -84,7 +84,7 @@ agent_framework/
   - `default_ttl: timedelta | None` — forwarded to the server as `params.task.ttl` (milliseconds). When `None`, the server's default applies.
   - `cancel_remote_task_on_local_cancellation: bool = True` — on local `CancelledError`, spawn a best-effort `tasks/cancel` before re-raising.
 - **Permissive fallback**: servers that ignore the augmentation (return `CallToolResult` directly) or reject the unknown `task` field with `METHOD_NOT_FOUND` / `INVALID_PARAMS` fall back to the plain `session.call_tool(...)` path so legacy servers keep working.
-- **Phase-aware reconnect**: a dropped connection before a `task_id` is known retries the augmented `tools/call`; once a `task_id` exists, only `tasks/get` / `tasks/result` are retried for the same id so reconnects never create duplicate remote tasks.
+- **Phase-aware reconnect**: a dropped connection before a `task_id` is known raises `ToolExecutionException("connection lost; task state unknown")` without re-issuing the augmented `tools/call`, so a server that accepted the request but lost the response cannot be made to start the same operation twice; once a `task_id` exists, `tasks/get` / `tasks/result` reconnect once and retry against the same id, which is safe because the request references the existing operation.
 
 ### File Access Harness (`_harness/_file_access.py`)
 

python/packages/core/agent_framework/_mcp.py

@@ -1344,7 +1344,7 @@ class MCPTool:
                     inner_exception=ex,
                 ) from ex
 
-    def _effective_task_options(self) -> "MCPTaskOptions":
+    def _effective_task_options(self) -> MCPTaskOptions:
         """Return the effective MCPTaskOptions, lazily constructing defaults on first use.
 
         Defers the implicit ``MCPTaskOptions()`` so the experimental warning only
@@ -1358,12 +1358,12 @@ class MCPTool:
         return self._task_options_default
 
     @property
-    def task_options(self) -> "MCPTaskOptions":
+    def task_options(self) -> MCPTaskOptions:
         """The effective MCPTaskOptions for this tool (lazy defaults)."""
         return self._effective_task_options()
 
     @task_options.setter
-    def task_options(self, value: "MCPTaskOptions | None") -> None:
+    def task_options(self, value: MCPTaskOptions | None) -> None:
         self._task_options_explicit = value
         self._task_options_default = None
 
@@ -1691,7 +1691,7 @@ class MCPTool:
             await asyncio.sleep(self._compute_poll_delay(snapshot.pollInterval).total_seconds())
 
     @staticmethod
-    def _coerce_get_task_result(lenient: "types.Result", task_id: str) -> "types.GetTaskResult":
+    def _coerce_get_task_result(lenient: types.Result, task_id: str) -> types.GetTaskResult:
         """Coerce a lenient Result into GetTaskResult, defaulting ``ttl`` when absent."""
         from mcp import types