{
  "slug": "stagehand",
  "name": "Stagehand",
  "url": "https://stagehand.dev",
  "cat": "review",
  "group": "ship",
  "tagline": "AI browser automation SDK",
  "price": "$0 forever, MIT open source; runs locally free, or against Browserbase from $20/mo",
  "tier": "free",
  "pick": false,
  "what": "Browser agent SDK from Browserbase built on Playwright, adding natural-language primitives: <code>act</code> to perform a step, <code>observe</code> to find actionable elements and <code>extract</code> to pull schema-validated data. Actions self-heal across markup changes, and you can drop back to plain Playwright code whenever determinism matters.",
  "why": "Selector-based tests shatter every time an AI rewrites your markup. Describing the step in English instead of pinning a DOM node keeps flows passing across cosmetic refactors you never asked for.",
  "warn": "Every natural-language step is an LLM call, so runs are slower, cost tokens and are less deterministic than plain Playwright. Cache resolved actions once a flow stabilises.",
  "install": "npm install @browserbasehq/stagehand",
  "category": {
    "key": "review",
    "name": "Code review, testing & security",
    "desc": "The single biggest risk in vibe coding is shipping AI code nobody read. This is the fix.",
    "group": "ship",
    "groupName": "Ship it & keep it alive"
  },
  "freshness": null,
  "health": {
    "ok": true,
    "status_code": 200,
    "final_url": "https://www.stagehand.dev/",
    "checked_at": "2026-08-19T00:37:41.862+00:00"
  },
  "corrected_at": null,
  "upstream_changed_at": null
}