API keys

The console at console.nufi.me is where you manage API keys for your own code — scripts, internal tools, integrations. You only need this page if you call NUFI from your own software. If you only use the app, ignore everything here.

You sign in at the app first; the console reads the same login. No second password. Open it from the app's avatar menu (bottom-left) → Console, or go straight to console.nufi.me.

The top navigation has three tabs — Profile, Usage, and API keys.

What the console shows

Profile (/)

• Your name and email.
• Your monthly budget and how much you have used so far.
• A 7-day usage chart.
• Your most-used API keys.

API keys

• A table of every key you have generated.
• Each row shows the key's nickname, when it was last used, its remaining budget, and a masked preview of the token.
• Buttons to generate new keys and revoke ones you no longer need.

Usage (/usage)

• Charts for the last 7, 30, or 90 days.
• Spend broken down by model.
• A table of your most recent requests.

Generate a key

If you lose a key, just revoke it and generate a new one.

Use a key

The key works with anything that speaks the OpenAI API format.

From the command line

curl https://api.nufi.me/v1/chat/completions \
  -H "Authorization: Bearer sk-nufi-…" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen2.5-7b",
    "messages": [{"role": "user", "content": "hello"}]
  }'

From Python

from openai import OpenAI

client = OpenAI(
    base_url="https://api.nufi.me/v1",
    api_key="sk-nufi-…",
)
print(client.chat.completions.create(
    model="qwen2.5-7b",
    messages=[{"role": "user", "content": "hello"}],
).choices[0].message.content)

From Node / TypeScript

import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'https://api.nufi.me/v1',
  apiKey: process.env.NUFI_API_KEY,
});

const reply = await client.chat.completions.create({
  model: 'qwen2.5-7b',
  messages: [{ role: 'user', content: 'hello' }],
});

console.log(reply.choices[0].message.content);

All the usual OpenAI parameters work — stream: true for streaming replies, tools: [...] for function calling, temperature, max_tokens, etc.

Revoke a key

Open the API keys tab, find the row, click **· · · ** → Revoke. The key stops working within seconds. Revoking cannot be undone — but you can always generate a fresh one with the same nickname.

Always revoke a key when:

• A laptop with the key on it is lost.
• A teammate who had the key leaves.
• You suspect the key has been exposed in a public repo or log.

Budgets and limits

Two layers protect you:

1. Account budget — your overall cap, set by your admin. Every key you issue counts against this.
2. Per-key budget — you can make individual keys tighter (e.g. a service key gets $5/month even if your account has $100).

When a budget runs out, calls return a 429 Budget reached error. Either wait for the next cycle, ask your admin to raise the account budget, or revoke and re-issue a key with a higher cap.

See exactly what your code did

Every call you make is logged with your account. The console's Usage page shows the last 50 calls — model, time, tokens spent. For deeper investigation (the actual prompt, the full reply, why a call was slow), your admin has access to a more detailed trace viewer.

First-time setup

The first time you open the console, your account is set up automatically with your organisation's default budget and limits. If you ever see "user not found" errors from api.nufi.me, just open the console once — that fixes it.

Best practices

• One key per app. Easier to revoke when you decommission the app.
• Never commit keys to git. Even private repos eventually leak. Use a secret manager or environment variable.
• Rotate every 90 days. Even without a known breach. Generate a new key, swap it in, revoke the old one.
• Set tight per-key budgets for unattended scripts. A bug should not drain your whole month.