Share your model with keys and limits

Put it in front of a model server you run, give each person or app its own key, and see who used what. Conversations are never stored.

Get it on GitHub How to start

The dashboard overview with token totals, requests, errors, speed, a chart of the last 24 hours and the busiest keys — **See how it's being used**Requests, tokens and speed for the last hour, day, week or month.

The keys page listing seven keys with their limits, status and last use — **A key for every app and person**Each key has its own expiry and its own limits, and shows how much of them it is using.

A new key shown once, with a copy button and a note that it will not be shown again — **Shown once, then kept safe**A new key is shown once. Quecto keeps only a scrambled copy it can check against.

The logs page with one row per request — **Know who used what**A line for every request, with its key, size and timing, never its content.

The settings page with server, logs and access settings, and a health section — **Change it while it runs**Point it at a different server or keep logs longer, and check that your model is answering.

How to start

You need Docker and a model server it can reach, such as llama.cpp, vLLM or Ollama.

Run it with Docker

Make an admin password, keep it somewhere safe, then start it.

git clone https://github.com/jesdga95/quecto-proxy && cd quecto-proxy
echo "QUECTO_ADMIN_PASSWORD=$(openssl rand -base64 24)" > .env && cat .env
docker compose up -d --build

Connect an app

http://your-host:11312/v1

Free and open source

Quecto Proxy is MIT licensed. The code and the full guide are on GitHub.

View on GitHub Read the guide See Quecto Harness