Editorial method · 22 October 2026

How we test browser battles

A fan battle can have the same character as the official game while using different health, keys, menus or attack timing. Our refreshed guides for Lancer, Asgore and Mettaton EX distinguish what we played from what we found in source or read in a creator's instructions. This page states the coverage of that 22 October check, not a claim that every battle has been cleared.

Identify the package before writing the advice

We downloaded the current HTML, packaged runtime and project archive from the same game host used by the site's player. For Scratch packages we extracted project.json and checked scripts, initializers and key handling. We also checked the current original-project metadata for author attribution. An upstream title or share date alone does not prove an exact version match.

The identifiers below are SHA-256 fingerprints of the hosted project source, not release names supplied by the creators. A changed byte changes the fingerprint. The host URLs are mutable, so a guide needs another check if the package is replaced. A later upstream edit does not automatically mean the hosted package changed too.

• sei26 — original project; project.json: 01ce15d1d31aa1b9207949d66d1f44a2a592df43780736c1260c8e67124d3885.
• jumpy8029 — original project; project.json: d7d17549656f4f157cb6e63a020ba4c43ea3a5edf746dd0eb936a93a09c9b303.
• PainInCode — original project; project.json: 59e373f17de0833409e15187d74e92f66b2d41e412167413f4928042c25b3aa2.

What the hands-on session covered

The browser reported Chrome 154 on macOS. Games were played at a 960x720 viewport using normal clicks and held/released keyboard inputs. We read runtime values to cross-check counters, but did not edit HP, teleport the SOUL, unlock turns, replace game scripts, or manufacture a victory. Keyboard automation is still a bounded practice session, not proof of a human no-hit clear or a broad device certification.

• Lancer: launch/menu, target and timed FIGHT, dialogue advance, the opening response and a subsequent DEFEND turn. We observed 90/110 party starting HP, Lancer falling from 500 to 325, and TP moving from 4 to 20 on DEFEND. The second-turn image shows Susie taking a hit despite Kris's shield.
• Asgore: launch, three-command row, a late meter attempt, a later 149-point hit and an early fire turn. The captured counters distinguish Asgore's 20000 starting HP from the player's 20. No victory, full item route or later spear sequence was completed.
• Mettaton EX: title-screen Z, opening menu, FIGHT, speech advance, first leg attack, directional input and X-slowed movement. A separate key check found that X, but not Shift, set this package's slow-input flag. No completed essay or final ratings outcome was tested.

How the screenshots are used

Six dated browser captures appear with captions in those three guides. They show the real canvas during the steps described, including ordinary damage and incomplete attempts. WebP and AVIF versions are compressed alternatives to the PNG captures; they are not AI-generated reconstructions. A screenshot proves the visible state at that moment, not every branch, the next turn, or the best possible result. We do not reuse a full-health opening image as evidence that the complete turn was hitless.

Source evidence is not playtest evidence

Finding an item handler or a quiz routine establishes that code is present. It does not establish that we completed it, that every input works, or that all ending conditions are reachable. The refreshed pages label those limits. Creator instructions can also disagree with a package: Mettaton's advertised Shift shortcut is one concrete reason we checked the actual key mapping rather than copying a control list unchanged.

A separate file comparison found identical Knight and Black Knife game archives, source and runtimes. We did not perform a new Knight playthrough. The Knight overview and weapon reference are two editorial entry points to the same current game, not evidence of two distinct difficulty builds.

Compatibility, tools and corrections

We checked responsive page layout at 320x740, 390x844, 844x390 and 1280x800, including footer reachability and the scrolling boss menu. Those were emulated viewport checks, not full Android, iPhone or Safari game runs. The text-box tool was tested separately for local-only input, pagination and export dimensions. Clipboard support remains browser-dependent.

Read the bounded build comparison and browser game help for practical consequences. Report a correction through Contact with the page, build and observed step. These checks verify content behavior, not ownership or permission to reuse third-party games. Older guides outside the three named here are not represented as newly play-tested by this page.