Upgrade rmcp to 3.3.0 and activate cancellation regression

Cargo.toml

@@ -24,7 +24,7 @@ repository = "https://github.com/contextforge-org/contextforge-data-plane"
 # Keep dependencies here only when at least two workspace members inherit them.
 contextforge-data-plane-cpex = { path = "./crates/contextforge-data-plane-cpex" }
 contextforge-data-plane-apis = { path = "./crates/contextforge-data-plane-apis"}
-rmcp = { version = "3.2.0", default-features = false, features = [
+rmcp = { version = "3.3.0", default-features = false, features = [
     "server",
     "client",
     "transport-streamable-http-server",

_context/wiki/testing.md

@@ -51,14 +51,18 @@ backends without recompiling a shared support tree for every feature file.
 | `gateway/{tools,prompts,resources,subscriptions}.rs` | Active routed operations and exact routing failures. |
 | `gateway/plugins.rs` | Gateway-owned CPEX ordering, mutation, denial, progress, and prompt seams using deterministic recording plugins. Resource coverage includes direct and aliased URIs, text/blob conversion, canonical pre-hook URIs, published-target rewrites, rejection of unpublished targets, metadata preservation, and pre/post denial. Concrete plugin behavior stays in each plugin crate. |
 | `gateway/harness/` | Authentication, modern and compatibility clients, in-memory configuration, concrete mock backends, and owned server fixtures. |
-| `gateway/future_contracts/` | Deferred fanout, pagination, TLS, completions, subscriptions, and cancellation contracts. |
+| `gateway/future_contracts/` | Deferred fanout, pagination, TLS, completions, and subscriptions contracts. |
+
+The active plugin suite verifies modern downstream cancellation reaches the
+backend, including cancellation before the response stream starts. The regression
+uses a three-second timeout so a cancellation deadlock fails promptly.
 
 `TestServer` binds `127.0.0.1:0` before spawning, uses cooperative
 cancellation, and has a `Drop` fallback. `GatewayFixture` owns the gateway and
 all backend servers. Tests should request the minimum topology: one virtual host
 and one backend by default, with extra backends declared explicitly by the case.
 
-The workspace currently keeps 13 ignored tests: 11 library future contracts
+The workspace currently keeps 12 ignored tests: 10 library future contracts
 and two real-process Redis/binary E2E tests. Ignored tests are not dead tests:
 keep them compiling, keep their intended assertions, give each a concrete
 blocker reason, and list them with:

crates/contextforge-data-plane-lib/tests/gateway/future_contracts/mod.rs

@@ -1,6 +1,5 @@
 mod completions;
 mod pagination;
-mod plugins;
 mod prompts;
 mod resource_templates;
 mod subscriptions;

crates/contextforge-data-plane-lib/tests/gateway/plugins.rs

@@ -610,6 +610,35 @@ async fn post_hook_deny_drops_progress_notifications_without_failing_call() {
     assert!(progress.lock().expect("progress lock poisoned").is_empty());
 }
 
+#[tokio::test]
+async fn downstream_cancellation_is_relayed_to_backend() {
+    tokio::time::timeout(std::time::Duration::from_secs(3), assert_downstream_cancellation())
+        .await
+        .expect("downstream cancellation relay completes within three seconds");
+}
+
+async fn assert_downstream_cancellation() {
+    let gateway = start_gateway(TEST_USER_ID, true, Arc::new(CpexRuntimeRegistry::default())).await;
+    let service = crate::harness::connect_modern_client(
+        gateway.gateway_url(),
+        crate::harness::create_client(TEST_USER_ID),
+        crate::harness::modern_client_info(),
+    )
+    .await;
+
+    let handle = service
+        .send_cancellable_request(
+            ClientRequest::CallToolRequest(Request::new(CallToolRequestParams::new("wait_for_cancellation"))),
+            PeerRequestOptions::no_options(),
+        )
+        .await
+        .expect("wait_for_cancellation request is sent");
+    wait_for_event_count(&gateway.backend_state.calls, 1).await;
+
+    handle.cancel(Some("client gave up".to_owned())).await.expect("cancellation is sent");
+    wait_for_event_count(&gateway.backend_state.cancellations, 1).await;
+}
+
 #[tokio::test(flavor = "multi_thread", worker_threads = 1)]
 async fn post_hook_can_return_raw_cmf_result_content() {
     let plugin = Arc::new(TestPlugin::new("post", vec![cmf_hook_names::TOOL_POST_INVOKE]).with_raw_post_rewrite());