browser-fetchlisted

# Browser Fetch Fallback fetcher for pages a normal HTTP fetch can't get past. Renders the page in a real (headless) Chrome via `undetected-chromedriver`, which defeats most headless-browser fingerprinting checks that trigger 403s. ## When to use As a fallback after `WebFetch` has already failed **or returned unusable content**, in either of these cases: - **Anti-bot blocking** — a 403, a Cloudflare/PerimeterX "checking your browser" page, or content that's clearly a bot-challenge shell rather than the real page. - **JavaScript-rendered pages** — WebFetch returns empty, truncated, or an obvious shell because the page builds its content client-side (SPAs, Swagger/ReDoc API docs, dashboards, admin UIs). WebFetch doesn't run JS; a real browser does, so the actual content appears. If WebFetch gives you a near-empty page or a small model's vague paraphrase of a docs site, this is the fix. Don't use this as a first attempt; it launches a real browser and is much slower than WebFetch. This is a **fetch-only** tool: it loads the URL, waits for JS/bot challenges to settle, and extracts the resulting text. It does not click, fill forms, or otherwise interact with the page. It will **not** help if the failure is actually geo-blocking or a login wall — those aren't bot-detection problems and a browser alone doesn't solve them. ## One-time setup (per machine) ```bash cd <this skill's directory> python3 -m venv .venv .venv/bin/pip install -r requirements.txt ``` Req