jackwener/OpenCLI · error · TimeoutError

No ChatGPT response appeared before timeout. Re-run with a h

Error message

No ChatGPT response appeared before timeout. Re-run with a higher --timeout if it is still generating.

What it means

TimeoutError raised when waiting for ChatGPT's answer: the ask loop watches the response text for stability (generation finished) but no response text ever appeared before timeoutSeconds elapsed.

Source

Thrown at clis/chatgpt/utils.js:2238

        if (await isGenerating(page)) {
            stableCount = 0;
            continue;
        }

        const messages = await getVisibleMessages(page, { textOnly: true });
        const candidate = findLatestNewAssistantResponse(messages, prompt, baselinePairCounts);
        if (!candidate || candidate === String(prompt || '').trim()) continue;

        if (candidate === lastText) {
            stableCount += 1;
            if (stableCount >= 2) return candidate;
        } else {
            lastText = candidate;
            stableCount = 0;
        }
    }

    throw new TimeoutError(
        'chatgpt ask',
        timeoutSeconds,
        'No ChatGPT response appeared before timeout. Re-run with a higher --timeout if it is still generating.',
    );
}

export async function getConversationList(page) {
    // ensureOnChatGPT already waits for the composer selector after navigation,
    // so the previous standalone 2 s settle is redundant.
    await ensureOnChatGPT(page);

    const openSidebar = requireBooleanEvaluateResult(unwrapEvaluateResult(await page.evaluate(`(() => {
        const button = Array.from(document.querySelectorAll('button'))
            .find((node) => /open sidebar/i.test(node.getAttribute('aria-label') || ''));
        if (button instanceof HTMLElement) {
            button.click();
            return true;
        }

View on GitHub (pinned to 49907e53dc)

Solutions

Re-run with a higher --timeout as the message suggests
Check the browser/conversation for rate-limit or error banners and retry after backoff
Verify the session is logged in and not blocked (ensureChatGPTLogin)
Split very large prompts or move long-running tasks to the deep-research flow with its own timeout

Example fix

// before
await chatgptAsk(page, hugePrompt);              // default timeout
// after
await chatgptAsk(page, hugePrompt, { timeout: 900 }); // 15 min

Defensive patterns

Strategy: retry

Try / catch

try {
  await waitForChatGPTAskResult(page, { timeoutSeconds: 120 });
} catch (e) {
  if (e instanceof TimeoutError && e.operation === 'chatgpt ask') {
    await page.sleep(30); // backoff
    // retry with timeoutSeconds: 600, after checking for rate-limit banners
  } else throw e;
}

Prevention

Set --timeout generously for long generations or tool-using prompts
Detect rate-limit/capacity banners and back off before retrying
Verify login state before the ask so a dead session doesn't burn the whole timeout

When it happens

Trigger: chatgpt ask wait loop polled until timeout with candidate text always empty/unchanged-empty — the model never started streaming a reply (still 'generating', refused, or an error state in the composer).

Common situations: Very long generation exceeding the default --timeout; ChatGPT rate-limit or capacity banner instead of a reply; prompt triggered a tool/deep-research that takes far longer; session degraded so no response renders.

Understand the failure class

Timeouts: ETIMEDOUT, deadlines, and hung requests — what actually expires when a request times out.

Related errors

AI-assisted analysis of jackwener/OpenCLI@49907e53dc (2026-08-29). Data as JSON: /api/errors/8da2f84d96596fda. Report an issue: GitHub.