How it works
Three steps to an honest map.
Connect your project
Point Mapparoo at the GitHub repository (read-only) and the planning documents — Kiro specs, Spec Kit artifacts, a design-doc set, or a folder of markdown notes. Any reasonable plan works; there's no format to learn.
Two independent reads
One analyst reads only the plan. Another reads only the code. Neither sees the other's work — then a separate judgment compares the two, and a verification pass tries to prove findings wrong. The guide says which ones it examined. No AI grading its own homework.
Read your map, act on it
An interactive guide you can understand in one sitting. Every gap arrives as a decision — what happened, what it threatens, what to do next — with a ready-to-paste fix brief for whoever writes your code, and a re-check to confirm the change actually landed.
And when it's time to show a cofounder, client, or investor: share a stakeholder edition — a clean, daylight-readable copy of your guide that needs no login and no explanation. Private by default, shared only when you say so.
Why trust it
Built to be honest, even when it stings.
A wrong finding is worse than no finding. Every claim Mapparoo makes is limited to what its evidence supports — and it tells you which evidence, at which strength.
No scores. Ever.
You will never see "your app is 78% complete." Percentages imply a certainty nobody has. You get findings with evidence, not a number that feels precise and means nothing.
Every "verified" names its evidence
"Verified" always says what supports it — for example, your declared build and tests completed in our audit environment. Passing tests are never presented as proof your product is ready for customers; Mapparoo will not overclaim what the evidence establishes.
"Unverified" is not an alarm
It means no evidence yet, either way — an ordinary, honest state, shown in calm gray. A tool that panics about everything is as useless as one that praises everything.
What we didn't examine, stated up front
Every guide opens with a coverage statement: what was read, what was analyzed, what was out of reach. Reduced coverage is displayed, never silently absorbed.
Judgments are labeled as judgments. Evidence is pinned to files and commits. Unknowns are shown, not smoothed over.
What "independent" means
Independent means independent analysis — not a different vendor. Mapparoo runs on Anthropic’s Claude models, the same family many founders build with. What makes the audit independent is how it is built, not who made the model. Your documents and your repository are read separately, with nothing shared between the two readings. The stages that judge and verify see only the validated results of those readings — never your raw documents, never your code, and never any session of the agent that wrote it. The verification pass tries to prove findings wrong, and the guide says which findings it examined. And every run records which model and prompt version each stage used. So no model grades work it produced, in a context it produced it in.