You’re judged on how you use AI, not whether you use it
Final code output is a weak signal on its own. Two candidates can ship the same diff through completely different processes — one who understood every line and verified it, one who pasted whatever the model produced. Promptster gives the hiring team the evidence trail to tell those apart: how you decompose a problem, what you verify, where you push back on the AI, and why you made the tradeoffs you did. Use AI as much or as little as you like. There’s no penalty for leaning on it heavily and no bonus for avoiding it. The signal is judgment — the quality of your direction and your verification.What the experience is like
You work the way you normally would:1
Install and start
Install the CLI and run
promptster start PST-XXXX-XXXX.2
Build
Open Claude Code, Codex, or Cursor from your assessment workspace and solve the task.
3
Explain your decisions
When you make a meaningful choice, run
/explain (in Claude Code) or promptster explain to record your reasoning in your own words.4
Submit
Run
promptster done. You get a link to your own session replay.Most assessments run in managed mode: your AI tool is routed through Promptster, the hiring team covers the cost, not you, and you don’t need a paid plan or to log out of a personal one. Some assessments have you use your own subscription instead —
promptster start sets up whichever the assessment uses. See Signing in to the AI.Key concepts
What is and isn’t recorded
Promptster only records activity inside your assessment workspace and your AI coding tool. It does not record keystrokes, screen contents, your clipboard, browser activity, environment variables, or files outside the workspace. See Privacy & Data for the complete list, including the optional authorship-timing layer you can decline at consent.Ready to begin?
Quick Start
Install, redeem your key, and start your assessment in under 5 minutes.
Troubleshooting
Hooks not capturing? Auth errors? Fixes for every common snag.
