Skip to main content
Things it can’t do on its own: search X/Twitter or Xiaohongshu, read any webpage, pull Reddit threads, look up products on Amazon or Taobao, fetch live crypto and market data, read SEC filings, check the weather or find a hotel — all through one key, with no per-source signups.
One line, about a minute: curl -fsSL https://agentkey.app/install.sh | bash on macOS/Linux, or irm https://agentkey.app/install.ps1 | iex on Windows. It detects the agents on your machine, installs the AgentKey skill into each one and opens a browser tab to sign in. Other methods — plugin marketplaces, npx, a prompt your chat agent runs, desktop config — are on Installation.
No. Install with one command — or let your coding agent install and set it up for you. After that you just ask your agent in plain language.If you want to write code, use the Scenario Methods REST API instead: fixed schemas, fixed prices, no MCP client needed.
Yes. POST https://api.agentkey.app/v2/ops/{method} with your key as a bearer token. The catalog, every method’s schema and the OpenAPI document are public, so you can evaluate the contract before signing up. See the Scenario Quickstart.
Social: X/Twitter, Reddit, YouTube, Instagram, TikTok, LinkedIn, Xiaohongshu, Douyin, Kuaishou, Bilibili, Weibo, Zhihu, WeChat Official Accounts and WeChat Channels. E-commerce: Amazon, AliExpress, Shopee, TikTok Shop, Taobao, JD.com, 1688, Xianyu, Douyin Shop and Xiaohongshu Shop. The raw catalog reaches further still — browse it with find_tools(prefix="social"). See Capabilities.
AgentKey is subscription-based. Each plan includes a monthly credit allowance covering everything your agent uses. When it runs out, buy extra credits on a paid plan or upgrade; the Free plan never converts on its own. Every Scenario Method has one fixed price shown in the catalog; every catalog tool’s per-call cost is shown when you describe it. Plans and prices are on the pricing page; see Subscription & Billing for how credits work.
Many agents — including Claude Code, Cursor, Windsurf, Gemini CLI, Codex, Warp, OpenClaw, Claude Desktop, and more — plus any MCP- or Skills-compatible client. See Supported Agents for the full list.
You can — but Twitter’s developer access alone is ~$100/month plus OAuth setup, then repeat for search, scraping, Reddit, and more. AgentKey gives your agent all of it through one account, and Scenario Methods fail over to a second provider for you when one goes down.
Scenario Methods try the next eligible provider automatically, refund the failed attempt and tell you what happened in attempts. Catalog tools run the exact tool you name, so there your agent picks an alternative — most capabilities exist on more than one provider. See Provider Failover.
It discovers tools at runtime with find_tools. See Discover → Describe → Execute.
The skill checks for a new version on the first call of a session and offers to upgrade; npx skills update agentkey forces it. To remove everything, run curl -fsSL https://agentkey.app/uninstall.sh | bash (macOS/Linux) or irm https://agentkey.app/uninstall.ps1 | iex (Windows). Details on Installation.
Most agents never see a key: they sign in to AgentKey through the browser. Where you do use a key — a client without MCP OAuth, or the REST API — never paste it into a chat or commit it to source control; store it as an environment variable. See Authentication.
The hosted service never sees your local files, credentials or conversations. The skill sends anonymous usage telemetry — which agent it runs in, its version, upgrade outcomes — never queries or responses. Opt out with touch ~/.config/agentkey/telemetry-disabled, or install with --no-telemetry.