superagnt_

OAuth first. The endpoint is its own authorization server, so almost every client (Claude Code, the Claude app, ChatGPT, Codex, Cursor, VS Code, Gemini, Windsurf, Zed, Grok, OpenClaw) runs the browser consent flow on first connect and gets its own credential. No token is ever pasted or revealed, and the consent screen doubles as signup. A workspace bearer token from the dashboard is the fallback for clients that still need one. The block at the top of this page prints the exact snippet for whichever one you run.

No. There's a free trial instead, and it doesn't ask for a card: verify your email and you get $3 of data credits and Growth limits. After that it's Starter, Growth or Scale, from $49/mo. Every plan has every capability. They differ only in how much you get.

When you actually use it, or 48 hours after you sign up at the latest. The 7-day clock starts at your 25th billable data call, the first time you use your workspace database, when the $3 runs out, or 48 hours after sign-up, whichever comes first. Connecting a client and letting your agent look around the tools don't count. If the credits run out early, only data calls stop and everything else keeps working until day 7. Then everything pauses, nothing is deleted, and your data is kept for 30 days so you can pick a plan and carry on.

A credit is a cent. Data credits pay for per-call data spend: data API calls and lead enrichment. Your plan includes a batch of them every month. AI credits pay for the model tokens and runtime a deployed agent burns, and you top those up in the dashboard. They are separate wallets on purpose, so an agent that runs long on tokens cannot empty the balance your lookups draw on. Credits come off on success, and every billable response reports what the call took and what is left.

Your agent tells you. Every cap is hard, including the data credits your plan includes. When a call would cross one, the agent gets an upgrade offer naming the next plan, what it unlocks and a link for you to confirm. Nothing is metered past the cap, so there is no overage bill. Move to a smaller plan and anything over its caps pauses. Nothing gets deleted.

No. Plans cover the platform and the data. The agent you already run (Claude Code, Codex, Cursor) brings its own model, so that costs you nothing extra here. Agents you deploy on superagnt_ burn AI credits you top up in the dashboard; the trial starts you with $1 of AI credit for hosted agents. Building one with the Agnt Wizard costs no AI credits.

Yes. Each plan has a requests-per-minute ceiling (60, 300 or 600), shared by every key and every MCP connection in your organization. Over REST, going over returns a 429 with a retryAfter telling you how long to wait. Over MCP, your agent gets the same wait time plus an offer for the next plan. A bigger plan buys throughput as well as quantity.

Yes. Keys, database, files and memory are all workspace scoped. The Postgres database is provisioned per workspace with no connection string handed out, and memory never crosses a workspace boundary. Your data and your agent's memory are yours, and they are not used to train foundation models.

Yes, and that is the intended path. A fresh connection starts on a small default surface so the tool list stays readable. From there agnt_tools_search finds what else exists and agnt_tools_enable switches a family on, so the agent widens its own toolset mid task instead of waiting for you to tick a box. agnt_tools_disable takes it back off.

No. For most clients the wiring is one line you paste into a terminal or a config file, and for the agent clients there is a prompt you can hand to the agent so it does the wiring for you. After that you talk to your agent, not to us. If you would rather not run MCP at all, every tool is also a REST endpoint you can call with the same key.