Small, working Replit templates that call paid APIs through TaskFuel.
Each one opens in Replit with a single click, runs on your own key, and costs cents per use. No provider accounts, no per-API keys, no subscriptions.
Type one brief. It goes to four image models at once and you get four takes back, each labelled with the price it actually charged. Around 48 cents for the set.
Ask X something in plain English. Your agent writes the search query, and up to twenty posts come back with their full text, long ones included, plus like and reply counts and a link to each original. Under a cent a question, and no X developer account.
Type a sentence. Grok, ElevenLabs and Deepgram each read it out loud, and every read is labelled with what it actually charged. Under four cents for all three, a couple of seconds each, and the prices differ by more than ten times.
Describe an object, or bring a picture of one. You see the picture first, then get a textured 3D model to spin in the browser and download as a GLB. About 9 cents from a sentence, 5 cents from your own picture.
Every paid call goes to one endpoint. TaskFuel pays the provider's HTTP-402 charge from your prepaid balance and passes the response straight back:
await fetch("https://app.taskfuel.ai/v1/call", {
method: "POST",
headers: {
Authorization: `Bearer ${process.env.TASKFUEL_API_KEY}`,
"Content-Type": "application/json",
},
body: JSON.stringify({
url: "https://some-provider.example/endpoint",
method: "POST",
body: { /* the provider's own parameters */ },
maxAmountUsd: 0.25,
}),
});The response headers say what it cost (x-taskfuel-cost) and what is left
(x-taskfuel-balance). GET /v1/discover?q=... searches every provider in the
catalog, around 90 of them, covering search, market data, email, phone calls,
images and more.
Full guide: app.taskfuel.ai/building-apps.md
Get one at app.taskfuel.ai. The first $1 is on the house.
In Replit, put it in the Secrets tab as TASKFUEL_API_KEY rather than in a
file. Secrets are not copied when someone else imports your project.
These templates run on your own key, and a key can spend the whole balance with nobody watching at call time. Every showcase here sets two limits in code, and you should keep them when you adapt one:
- A per-call ceiling (
maxAmountUsd), so a single call cannot run away. - A daily budget that fails closed once it is reached.
If you make a project public and let strangers trigger paid calls, they are spending your balance. Keep the budget low, or have each visitor bring their own key.
One repo per showcase, because Replit's import URL takes a whole repository and
.replit names a single entrypoint. A repo holding several templates would
hand every importer all of them, with only one of them running.
So: create taskfuel/<showcase-name>, make it public, and add a section here
pointing at it.