How I use: Pythond, a persistent Python runtime and Cloakbrowser together to give agents a persistent stealth browser

[rangersui]

pythond + CloakBrowser: a persistent stealth browser environment for agents

I built pythond (pip install pythond), a persistent Python daemon where state survives across calls.

Combined with CloakBrowser, my AI agent can spin up a stealth browser once and keep it alive forever:

# one-time setup — browser launches and stays in the namespace
pysh run work "exec(open('cloak_browser.py', encoding='utf-8').read())"

# browse freely — state persists between calls
pysh run work "goto('https://protected-site.com')"
pysh run work "text()"
pysh run work "links(pattern='login')"
pysh run work "fill('#email', 'me@example.com')"
pysh run work "click('text=Sign in')"

# ... hours later, browser still alive, still logged in ...
pysh run work "goto('https://protected-site.com/dashboard')"
pysh run work "tables()"

No cold starts. No re-launching Chromium. No re-passing bot detection on every call. Stay logged in across sessions.

CloakBrowser handles the stealth. pythond handles the persistence. Together: a permanent stealth browser for AI agents.

Thanks to @Cloak-HQ for making this possible. Every intelligent agent deserves the right to browse the web.

Links

cloak_browser.py
pythond | pythond.sh