SOURCES
What this guide stands on.
Two kinds of source: the developer's own material, and our own sessions. There is no third kind on this site.
Primary
| Source | What we take from it | Type |
|---|---|---|
| Store page for Project Restoration | Premise, features, controls, tags, rating, update date | Store record, read 2026-10-31 |
| The game client itself | Everything measured: timings, costs, maps, dialogue | First-hand, dated per session |
| project-restoration.com | The developer's landing page | Checked 2026-10-31 — one client-rendered page, no sitemap |
Secondary, used only for orientation
Three sources tell us what is already written about the game without being usable as evidence: a Fandom wiki with 1,806 pages (our sample of seven ran 16–34 words each), Russian walkthrough sites that predate the browser port, and a long tail of portal pages that repeat the store blurb.
We read them to learn what players ask. We do not copy numbers from them, and where one disagrees with the client, the client wins and the disagreement gets written down.
How the wiki numbers were read
The community wiki was read through its own MediaWiki API on 31 October 2026, from a browser
rather than a script, because the host refuses non-browser clients outright. Two calls did the
work: list=allpages returned 2,108 page titles, and list=allcategories
plus per-category members returned 182 categories and their contents. Counts here are category
sizes, so a page filed twice is counted twice and a page nobody categorised is missing. That is
the trade: a complete name inventory with a known, stated margin of error, rather than a partial
one with none. Every read is dated on the page that uses it, because a category size is only true on
the day it was counted.
What we did not read
The wiki’s talk pages, its file metadata and its 51 update articles were left alone except for the date lines, so anything that only exists in a discussion thread is missing here. We did not cite the game’s Discord, any video, or any of the Russian- and Turkish-language pages that follow the original release, even though several look deeper than the English side, because none of them date their claims, which makes them leads rather than sources. And we did not ask the developer a single question: there is no press kit, no documentation site, and the official domain serving 3 KB of markup is not a place to send a list of forty questions.
What the site will not do with sources
No unattributed claims, no figures without a date, and no page that exists only to repeat a store description under a different title.