PrimeTime
Early-stage, open source

Keep your AI coding CLI warm before you sit down to code.

PrimeTime schedules a lightweight “primer” request for Claude Code, Codex, and other AI coding CLIs, so your provider's usage window has already started by the time your normal coding session begins.

How it works

1

Sign in and create your warmup repository

Sign in with GitHub, then click one button. PrimeTime installs itself and creates a small, private repository dedicated to warming up your AI coding CLI — no picking through your existing projects.

2

Hand off your provider session once, locally

Run the PrimeTime CLI on your own machine after logging into Codex (or another supported provider). It reads your local session and stores it directly in that repository's GitHub Actions secrets — it never passes through PrimeTime's servers.

3

Copy one workflow file in, on a schedule

Copy the provided GitHub Actions workflow into .github/workflows/. It runs on your schedule, sends one lightweight primer request through the official provider CLI, and writes the refreshed session back to your own secrets.

Your credentials never touch PrimeTime

Provider sessions live only in your own GitHub Actions secrets. PrimeTime's backend mints short-lived, repository-scoped GitHub tokens — it never receives, stores, or logs a Codex, Claude, or other provider credential.

Runs on your schedule

A timezone- and DST-aware scheduler figures out exactly when your primer should fire, based on your normal work hours and how much lead time your provider needs.

Prefer a fully self-hosted setup with no backend in your trust chain at all?

Self-host instructions