jackwener/OpenCLI · error · CommandExecutionError

Browser session required for medium posts

Error message

Browser session required for medium posts

What it means

loadMediumPosts scrapes Medium user pages through a browser (page.goto/wait/evaluate). It requires an active browser session; when page is null/undefined it throws CommandExecutionError 'Browser session required for medium posts' instead of attempting a headless HTTP fetch.

Source

Thrown at clis/medium/utils.js:13

import { CommandExecutionError } from '@jackwener/opencli/errors';
export function buildMediumTagUrl(topic) {
    return topic ? `https://medium.com/tag/${encodeURIComponent(topic)}` : 'https://medium.com/tag/technology';
}
export function buildMediumSearchUrl(keyword) {
    return `https://medium.com/search?q=${encodeURIComponent(keyword)}`;
}
export function buildMediumUserUrl(username) {
    return username.startsWith('@') ? `https://medium.com/${username}` : `https://medium.com/@${username}`;
}
export async function loadMediumPosts(page, url, limit) {
    if (!page)
        throw new CommandExecutionError('Browser session required for medium posts');
    await page.goto(url);
    await page.wait({ selector: 'article', timeout: 5 });
    const data = await page.evaluate(`
    (async () => {
      await new Promise((resolve) => setTimeout(resolve, 3000));

      const limit = ${Math.max(1, Math.min(limit, 50))};
      const normalize = (value) => (value || '').replace(/\\s+/g, ' ').trim();
      const posts = [];
      const seen = new Set();

      for (const article of Array.from(document.querySelectorAll('article'))) {
        try {
          const titleEl = article.querySelector('h2, h3, h1');
          const title = normalize(titleEl?.textContent);
          if (!title) continue;

          const linkEl = titleEl?.closest('a') || article.querySelector('a[href*="/@"], a[href*="/p/"]');

View on GitHub (pinned to 49907e53dc)

Solutions

1. Start the browser session/Browser Bridge before running the medium posts command (per the CLI's browser setup instructions).
2. Verify the page object is passed to loadMediumPosts — never call it with an undefined page in custom scripts.
3. Re-open the browser if a previous session crashed, then rerun the command.
4. If you need data without a browser, use the RSS-based medium tag command instead, which does not require a page.

Example fix

// before
await loadMediumPosts(null, 'https://medium.com/@user', 5); // throws
// after
const page = await browser.newPage();
await loadMediumPosts(page, 'https://medium.com/@user', 5);

if (!page) {
  throw new Error('Start the browser session/Browser Bridge before loading medium posts');
}

function hasBrowserPage(p) {
  return p != null && typeof p.goto === 'function' && typeof p.evaluate === 'function';
}

try {
  await loadMediumPosts(page, url, limit);
} catch (e) {
  if (/Browser session required/.test(e.message)) {
    const page = await startBrowserBridge();
    await loadMediumPosts(page, url, limit);
  } else throw e;
}

When it happens

Trigger: Calling loadMediumPosts(page=null, url, limit) or invoking the wrapping medium posts command without a browser session/bridge established (e.g. browser not started, Browser Bridge not connected, session closed before the call).

Common situations: Running the command in an environment without a logged-in/available browser; forgetting to initialize the browser adapter before calling the medium posts helper; a previously-open page that crashed or was closed leaving `page` falsy.

Related errors

• Browser session required for linkedin-learning trending
• Browser session required for gmail attachments
• Browser session required for gmail labels
• Browser session required for linkedin connect
• medium tag request failed: ${err?.message ?? err}

AI-assisted analysis of jackwener/OpenCLI@49907e53dc (2026-08-29). Data as JSON: /api/errors/2d6696ce5bf0dd98. Report an issue: GitHub.