Someone, a product manager, a designer, a founder-adjacent enthusiast, vibe-coded a prototype that worked well enough to win a budget. Now an engineering team inherits it, and the gap between demo and production system becomes their problem. This handover has the widest skill asymmetry of any on this site, in the unusual direction: the recipient knows far more than the builder.
Teams inheriting prototypes default to one of two moves. Adopt it wholesale, and inherit every unexercised path and unreviewed generated corner as latent production incidents. Or rewrite from scratch, and throw away months of validated product decisions because the code that embodies them cannot be trusted. Both defaults are really the same failure: nobody can tell, cheaply, which parts of the prototype are load bearing and which are agent enthusiasm. The decision the team actually needs to make, keep what, rewrite what, harden what, requires an evidence map nobody has.
Run the CLI on the prototype before the team takes it. The scan reads the repository and whatever agent history the builder’s machine holds, which for prototypes is often the only design documentation in existence: the entire reasoning lives in Claude Code or Cursor sessions. Verify executes install, build and whatever checks exist in a clean sandbox, establishing the baseline does it even run without the builder’s laptop. The passport that comes out gives the team the triage sheet: what is repository-confirmed, what was executed and passed, what only the agent ever claimed, and what nobody checked, each labelled per the truth model.
With the passport on the table, the inheritance meeting changes shape. The decision log preserves why the prototype does what it does, so product knowledge survives even where code does not. The contradicted claims and unresolved entries are the rewrite shortlist, pre-argued. The verified setup means the team runs the thing on day one instead of day eight. And the builder, who was dreading being cross-examined by engineers, hands over a document instead of defending a memory. Prototypes with passports get adopted incrementally and gratefully; prototypes without them get rewritten defensively.
If the prototype builder is the founder and the team is one engineer, that intimate version is the first technical hire. If the prototype came from an outside contractor, add the commercial close from contractor offboarding. And if the inheriting team wants the checks re-executed somewhere neutral before staking a roadmap on them, Mantyl Verifiedsigns an independent run for a one-off £19 to £49, which is cheap insurance under a quarter’s planning.