fix(agents): skip empty embedded prompts

CHANGELOG.md

@@ -26,6 +26,7 @@ Docs: https://docs.openclaw.ai
 - Control UI/Codex harness: emit native Codex app-server assistant and lifecycle completion events so live webchat runs stop spinning without needing a transcript reload fallback. (#70815) Thanks @lesaai.
 - Agents/sessions: persist the runtime-resolved context budget from embedded agent runs, so Codex GPT-5.5 sessions keep the catalog/runtime context cap instead of falling back to the generic 200k status value. Fixes #71294. Thanks @tud0r.
 - Agents/tools: fail runs before model submission when explicit tool allowlists resolve to no callable tools, preventing text-only hallucinated tool results for missing tools such as plugin commands that were not registered. Fixes #71292.
+- Agents/embedded: skip provider submission when an embedded run has no prompt, replay history, or prompt-local images, preventing empty OpenAI Responses requests from surfacing provider errors into user channels. Fixes #71130.
 - Providers/MiniMax: keep M2.7 chat model metadata text-only so image tool requests route through `MiniMax-VL-01` instead of the Anthropic-compatible chat endpoint. Fixes #71296. Thanks @ilker-cevikkaya.
 - Discord/replies: run `message_sending` plugin hooks for Discord reply delivery, including DM targets, so plugins can transform or cancel outbound Discord replies consistently with other channels. Fixes #59350. (#71094) Thanks @wei840222.
 - Control UI/commands: carry provider-owned thinking option ids/labels in session rows and defaults so fresh sessions show and accept dynamic modes such as `adaptive`, `xhigh`, and `max`. Fixes #71269. Thanks @Young-Khalil.

src/agents/pi-embedded-runner/run/attempt.empty-prompt.test.ts

@@ -0,0 +1,64 @@
+import type { AgentMessage } from "@mariozechner/pi-agent-core";
+import { afterEach, beforeEach, describe, expect, it, vi } from "vitest";
+import {
+  cleanupTempPaths,
+  createContextEngineAttemptRunner,
+  createContextEngineBootstrapAndAssemble,
+  resetEmbeddedAttemptHarness,
+} from "./attempt.spawn-workspace.test-support.js";
+
+describe("runEmbeddedAttempt empty prompt guard", () => {
+  const tempPaths: string[] = [];
+
+  beforeEach(() => {
+    resetEmbeddedAttemptHarness();
+  });
+
+  afterEach(async () => {
+    await cleanupTempPaths(tempPaths);
+    vi.restoreAllMocks();
+  });
+
+  it("skips provider submission when prompt, history, and images are empty", async () => {
+    const sessionPrompt = vi.fn(async () => {});
+    const { assemble } = createContextEngineBootstrapAndAssemble();
+
+    const result = await createContextEngineAttemptRunner({
+      contextEngine: { assemble },
+      sessionKey: "agent:main:guildchat:dm:empty-prompt",
+      tempPaths,
+      sessionMessages: [],
+      sessionPrompt,
+      attemptOverrides: {
+        prompt: "   ",
+      },
+    });
+
+    expect(sessionPrompt).not.toHaveBeenCalled();
+    expect(result.promptError).toBeNull();
+    expect(result.finalPromptText).toBeUndefined();
+    expect(result.messagesSnapshot).toEqual([]);
+    expect(result.assistantTexts).toEqual([]);
+  });
+
+  it("still submits a blank prompt when replay history has content", async () => {
+    const sessionPrompt = vi.fn(async () => {});
+    const { assemble } = createContextEngineBootstrapAndAssemble();
+    const sessionMessages = [
+      { role: "user", content: "previous turn", timestamp: 1 },
+    ] as AgentMessage[];
+
+    await createContextEngineAttemptRunner({
+      contextEngine: { assemble },
+      sessionKey: "agent:main:guildchat:dm:empty-prompt-with-history",
+      tempPaths,
+      sessionMessages,
+      sessionPrompt,
+      attemptOverrides: {
+        prompt: "   ",
+      },
+    });
+
+    expect(sessionPrompt).toHaveBeenCalledTimes(1);
+  });
+});

src/agents/pi-embedded-runner/run/attempt.spawn-workspace.test-support.ts

@@ -236,6 +236,7 @@ vi.mock("../../../infra/machine-name.js", () => ({
 }));
 
 vi.mock("../../../infra/net/undici-global-dispatcher.js", () => ({
+  DEFAULT_UNDICI_STREAM_TIMEOUT_MS: 120_000,
   ensureGlobalUndiciEnvProxyDispatcher: (...args: unknown[]) =>
     hoisted.ensureGlobalUndiciEnvProxyDispatcherMock(...args),
   ensureGlobalUndiciStreamTimeouts: (...args: unknown[]) =>
@@ -281,6 +282,8 @@ vi.mock("../../pi-project-settings.js", () => ({
   createPreparedEmbeddedPiSettingsManager: () => ({
     getCompactionReserveTokens: () => 0,
     getCompactionKeepRecentTokens: () => 40_000,
+    getGlobalSettings: () => ({}),
+    getProjectSettings: () => ({}),
     applyOverrides: () => {},
     setCompactionEnabled: () => {},
   }),
@@ -302,6 +305,7 @@ vi.mock("../extensions.js", () => ({
 }));
 
 vi.mock("../replay-history.js", () => ({
+  normalizeAssistantReplayContent: <T>(messages: T) => messages,
   sanitizeSessionHistory: async ({ messages }: { messages: unknown[] }) => messages,
   validateReplayTurns: async ({ messages }: { messages: unknown[] }) => messages,
 }));

src/agents/pi-embedded-runner/run/attempt.ts

@@ -453,6 +453,14 @@ function summarizeSessionContext(messages: AgentMessage[]): {
   };
 }
 
+function hasPromptSubmissionContent(params: {
+  prompt: string;
+  messages: readonly AgentMessage[];
+  imageCount: number;
+}): boolean {
+  return params.prompt.trim().length > 0 || params.messages.length > 0 || params.imageCount > 0;
+}
+
 export function applyEmbeddedAttemptToolsAllow<T extends { name: string }>(
   tools: T[],
   toolsAllow?: string[],
@@ -2373,6 +2381,28 @@ export async function runEmbeddedAttempt(
             transcriptLeafId,
           });
 
+          if (
+            !skipPromptSubmission &&
+            !hasPromptSubmissionContent({
+              prompt: effectivePrompt,
+              messages: activeSession.messages,
+              imageCount: imageResult.images.length,
+            })
+          ) {
+            skipPromptSubmission = true;
+            log.info(
+              `embedded run prompt skipped: empty prompt/history/images ` +
+                `runId=${params.runId} sessionId=${params.sessionId} trigger=${params.trigger} ` +
+                `provider=${params.provider}/${params.modelId}`,
+            );
+            trajectoryRecorder?.recordEvent("prompt.skipped", {
+              reason: "empty_prompt_history_images",
+              prompt: effectivePrompt,
+              messages: activeSession.messages,
+              imagesCount: imageResult.images.length,
+            });
+          }
+
           // Diagnostic: log context sizes before prompt to help debug early overflow errors.
           if (log.isEnabled("debug")) {
             const msgCount = activeSession.messages.length;