Keep question-tool routing visible before deferred discovery

Retain one sentence routing questions to Paperclip while the detailed usage lives in the tool documentation. Account for the child completion wake in the continuation fixture turn estimate.

Co-Authored-By: Paperclip <noreply@paperclip.ing>

server/src/services/native-runtime/native-execution-input.test.ts

@@ -461,6 +461,8 @@ describe("native execution input external-chat framing", () => {
       runtimeContext: nativeRuntimeContextFixture(),
     });
     expect(input.task.prompt).not.toContain("## Questions that need a user response");
+    expect(input.task.prompt).toContain("Use Paperclip's request_human_input for user questions");
+    expect(input.task.prompt).not.toContain("payload.questionSet");
   });
 
 });

server/src/services/native-runtime/native-execution-input.ts

@@ -156,6 +156,9 @@ export function buildNativeExecutionInput(input: {
     isPaperclipExternalChatQuestionResponseTurn(wakePayload);
   const taskPrompt = [
     wakePrompt,
+    // Provider-local elicitation can wait without creating a Paperclip card.
+    // Keep routing visible before deferred tool discovery; usage belongs in the tool schema.
+    "Use Paperclip's request_human_input for user questions; provider-local question tools do not reach the Paperclip UI.",
     externalChatTurn && wake?.externalChatProvider === "github"
       ? NATIVE_GITHUB_ATTACHMENT_RECOVERY_GUIDANCE
       : "",

tests/runner-e2e/README.md

@@ -161,7 +161,7 @@ Both suites save and restore experimental settings. Browser E2E always starts a
 throwaway instance; never point the authenticated suite at the running demo.
 Missing provider credentials fail paid preflight and are not passing coverage.
 
-The default `--all` selection is 160 cells (137 local and 23 Daytona) and 340
+The default `--all` selection is 160 cells (137 local and 23 Daytona) and 344
 expected paid agent turns. The explicit-only everyday suite adds 30 catalog cells
 and is excluded from `--all`. Follow-up steps remain ordered within their cell; all other
 cells are independent. Narrow selectors are strongly recommended while

tests/runner-e2e/catalog.test.ts

@@ -68,7 +68,7 @@ describe("runner E2E catalog", () => {
         (total, execution) => total + execution.task.expectedRunCount,
         0,
       ),
-    ).toBe(340);
+    ).toBe(344);
     expect(
       runnerTasks.find((task) => task.id === "plan-revise-accept")
         ?.attemptTimeoutMs,

tests/runner-e2e/continuation-cases.ts

@@ -55,13 +55,15 @@ export const continuationTasks: readonly RunnerTaskFixture[] =
     groups: [],
     workMode: "standard",
     flow: "continuation",
-    expectedRunCount: [
-      "clarification-not-approval",
-      "revision-preserves-approval",
-      "completed-action-resume",
-    ].includes(id)
-      ? 3
-      : 2,
+    expectedRunCount:
+      id === "completed-action-resume"
+        ? 4
+        : [
+              "clarification-not-approval",
+              "revision-preserves-approval",
+            ].includes(id)
+          ? 3
+          : 2,
     attemptTimeoutMs: { local: 10 * 60_000, daytona: 10 * 60_000 },
     expectedTerminalState: { issue: "done", run: "succeeded" },
     buildTitle: (nonce) => `Continuation ${id} ${nonce}`,