Fast-path lookups — no credentials needed
A few of the many sites where common questions resolve in 1-4 seconds because we maintain structured indexes against them. Just phrase the question naturally; stablebrowse figures out the best data path.Hacker News
top, new, best, ask, show, jobs; story details; comments
Wikipedia
search, article, summary, sections, references, recent changes, random
Stack Overflow
search, question, answers, tags, users
Steam
search, app details, reviews, featured, categories
Yelp
search, business details, autocomplete
Amazon
search, product, deals, bestsellers, departments
Airbnb
search, listing, dates, guests, amenities
Craigslist
search by city + category
GitHub
repos, issues, PRs, code search
Google Flights
search by origin / destination / dates
LeetCode
problems, filters, difficulty, tags
Redfin
listings, properties, sale / rent
Starbucks
locations, hours, menu
Uber
fare estimates, ride types
Walmart
search, products, deals
Zillow
listings, properties, sale / rent
Available without end-user credentials
LinkedIn and YouTube work out of the box — no end-user cookies required. Just reference them in your task prompt and stablebrowse handles the lookup.search people and companies, profile lookups
YouTube
search, video details, comments, channels
Social platforms
Public data on these platforms works out of the box — stablebrowse supplies a default, unpersonalised set of session cookies automatically, so searches, trending, public profiles, and similar read-only queries just work without any setup on your end. Supply the end-user’s own session cookies via the credentials endpoint only when you want personalised answers: their home feed, their own profile, DMs, watch history, private content, etc. Results come back in 5-30 seconds either way.Twitter / X
Twitter / X
Cookies needed:
twitter_auth_token, twitter_ct0Capabilities: searching tweets, fetching a user’s timeline or replies, user profiles, home timeline, trending topics, following lists.How to get the cookies: log in to x.com in a browser → DevTools → Application → Cookies → copy auth_token and ct0.Reddit
Cookies needed:
reddit_session (only for authenticated features like fetching your own profile)Capabilities: browsing subreddits, searching posts, reading comment threads. Most features work without any cookies.TikTok
TikTok
Cookies needed for auth-gated features:
tiktok_session_id, tiktok_csrf_tokenCapabilities: hashtag pages, trending, user profiles, video detail, comments, search users, followers/following, your own profile. Public features (hashtag, trending, video, profile) work without cookies.Instagram
Cookies needed:
instagram_session_id, instagram_csrf_token, instagram_ds_user_idCapabilities: profiles, full profiles, searching reels, hashtag posts, and more.Everything else on the web
Any site not listed above runs through the full browser agent — the system reads the page, plans its next action, clicks, scrolls, and extracts. Works for most of the public web: blogs, news, dashboards, one-off research. Typically 10-60 seconds per task. This is also the fallback when a site IS in the fast-path list but the specific question doesn’t fit a pre-indexed pattern.Credential storage
All credentials are encrypted at rest via AWS KMS. They’re attached to a specific(businessId, endUserId) pair and never returned in any GET response — the status endpoint only says “configured: true/false” per platform. Delete them at any time via DELETE /v1/end-users/{id}/credentials.