explore superbot

superbot

it uses your computer. you watch from your phone.

Install now

make it, don’t search for it.

it works for you.

the superbot composer with its computer-use menu open: off, ask before each action, or always

superbot keeping a task going on its own while you are away

everywhere you are.

make it yours.

superbot appearance settings: the accent row open with seven colors and a color pad, the whole app in pink

superbot appearance settings with the ash scheme picked: the whole app turns grey

superbot appearance settings with tokyo night picked from the popular palettes

superbot appearance settings: chat font size and density

and a lot more.

get superbot.

Install now

developers start at the api and the docs.

the relay

The API key for your Subscriptions

https://api.superbot.gg/v1

19,050requests in 30d
+$521saved in 30d
3apps on one key

The Relay app's Overview: the relay switched on with 635 requests and +$11.99 saved in 24 hours, the live call strip, the Endpoint box with https://api.superbot.gg/v1, Copy and Mint key, Your apps with Claude Code and Codex on and Cursor off, and the Plan limits for each subscription.

The Relay app's Team page: prompts, tokens, chats, spend and API equivalent for the week, the week's activity chart, and the People table with each person's state (Active, Near limit, Over limit), prompts, spend and last prompt.

The Relay app's Team spend page: spend $2,744, API equivalent $20,600, saved +$17,856 and 56% of the budget used, a spend against API equivalent chart for the week, and each person's month against their limit, with Set limit.

every call, one log

every app's calls, one live feed.
search and filter any of them.
omni and limit hits show as they land.

The Activity tab with the live log on: logging is on and the Live dot shows, then a table of eight calls with time, app, model, account, tokens, wait and status, one Omni account pill and one 429 row.

open any call

every span, timed.
tokens and cache reads.
what it would cost at API rates.

A call's details for a 200 Claude Code call with All fields open: the field grid shows cache read 19,800 and cost at API rates $0.0391, above a trace waterfall of waiting, then streaming. — a call that hit its limit

A call's details for a 429 call, compact: a red Rate limited headline with its 429 badge, a red callout saying this account has reached its five-hour limit, and the five key facts. — a call that hit its limit

your accounts

your plans and keys, one place.
live limits on every plan.
credentials are never shown.

The Accounts tab with its sidebar: a plans table with Claude folded to three accounts and its five-hour and seven-day limit bars, ChatGPT expanded to its Plus plan with its own limit bars, and GitHub with no limit yet; a note says credentials are never shown here, and a Models card groups the models by account.

built in

omni runs claude, tidy trims calls.
uploads ship compressed deltas.
refusals can go to superbot.

The Relay Overview's superbot relay card with its Options panel open: the On switch and 635 requests in 24h, then five settings switches. Tidy trims what each call loads, Omni runs claude (Claude Code 2.1.288), Compress uploads compresses requests where supported, and Delta uploads skips what the server already has, all on; Redirect refusals to Superbot, where Superbot answers when a model refuses, is off.

Install superbot relay

the api

your plans, one endpoint.

https://api.superbot.gg/v1 Mint key

mint a key
set two env vars
call superbot

API reference OpenAPI spec

OPENAI_BASE_URL=https://api.superbot.gg/v1
OPENAI_API_KEY=sbc_YOUR_KEY

curl https://api.superbot.gg/v1/chat/completions \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "superbot",
    "messages": [{ "role": "user", "content": "hi" }]
  }'

response

{
  "id": "chatcmpl-…",
  "object": "chat.completion",
  "model": "superbot",
  "choices": [{
    "index": 0,
    "message": {
      "role": "assistant",
      "content": "hi! what should I do?"
    },
    "finish_reason": "stop"
  }]
}

example response