One recommendation that looks great on paper and then no-shows or can't deliver burns your whole team's credibility. This scores a candidate on the three things that actually predict it — reliable, technically competent, resourceful — and gives you a defensible GO / NO-GO.
Grounded in 85 years of selection research: work samples & structured references are the top predictors; unstructured reference checks are near-noise. Runs entirely in your browser — no data leaves the page.
| Criterion | Instrument | Hard rule |
|---|---|---|
| Reliable | Structured reference from someone who worked with them | a single “went dark / no-show” ⇒ automatic NO-GO |
| Technical | 90-min observed work sample, one shared rubric | must clear the role bar; gates technical roles only |
| Resourceful | learn an unfamiliar tool in a day + teach it back | must ship something that runs and teach it |
GO only if every required axis clears on real evidence. A strong axis never rescues a weak one. This tool advises — the recommender still signs.
Use an anonymized id. Fill from the three instruments (reference form, work-sample score, teach-back score).
A vouch from a client team or your own team both count. One “went dark / no-show” is disqualifying.
Outcome over method — using AI / asking for help is fine; the bar is it works and they can teach it.
role: technical