Koalinku: The Living Isles — supported playtest guide

Build: Welcome Woods + Sanctuary Skills + Grove Guardians, 16 October 2026.

What is ready to test

A local, saved 3D island with building, creatures, a learning trail, two treasure puzzles, five additional world activities over two linked activity levels, companion abilities, a new Glowmoth visitor, on-screen controls, adjustable reading panels, optional sound and local-device read aloud. This is a playtest candidate, not an independently validated educational release.

No real children, parents or educators were recruited by the developer assistant. Automated checks, developer browser play and prepared test saves do not replace those sessions. Do not present the sample or development feedback as participant findings.

Before inviting someone

• An adult should first finish a short session on the intended device. Check that WebGL opens, movement works, the game saves and a backup can be restored.
• Use a separate browser profile/device for each independent participant, or export the previous world before resetting. Saves are shared by pages on the same local browser origin. Never reset an important world without a backup.
• Choose a broad reading profile, not a birthday. Choose Guided or Standard clues separately. Young readers can co-play with an adult. Check a local English voice is available if read aloud is needed; printed text remains when it is not.
• Agree participation with the adult supporter and player. Explain that this tests the game, not whether the child is good, clever or well behaved. They can skip a question, pause, stop, or decline a hug in the fictional story.
• Use no names, school names, birthdays, contact details, photographs or recordings in the feedback files. Do not post children’s world backups publicly: backups include their fictional game choices. Keep any optional session notes anonymous and under the facilitator’s control.

Prepared practice world

The ZIP includes playtest-worlds/Grove-Guardians-Practice.json. It is an authored setup with the earlier activity level finished, known companions and an open bridge, so a short session can focus on later puzzles. Export an important existing world first, then use Help → Restore backup to open the practice file. It contains no participant data. Reports automatically mark that a prepared world was used; record this when discussing results.

The localhost preview address works on this computer. A physical phone needs an approved review URL reachable from that device; localhost on a phone points to the phone itself. Static files for a test host are supplied in the ZIP. No public upload is included.

A 15–25 minute session

1. Ask the player to look around and find how to move. Avoid explaining the controls immediately. Note whether they discover keyboard, on-screen buttons, Interact and the Adventures book. Offer help if asked or if frustration is building.

2. Invite them to collect a material bundle and build a fern. Try both ground pointing and preview arrows + Place here. Check whether the placement message makes sense and whether touch buttons leave enough room to see the world.

3. Try one learning encounter and its practical action. Ask a neutral question: “What do you think might happen?” After feedback, ask “What could help next?” Accept different reasonable explanations. Do not coach for a high score.

4. Open Adventures. Choose pond cleanup, planting or the stepping trail. Can they understand the goal, follow the activity arrow, notice completion and locate a companion’s ability? For young readers, begin with the three-symbol trail. Older profiles use four or five steps, without a timer.

5. If there is time or a prepared facilitator world, test Grove Guardians. Rotate connected channels, call Brookfin, grow Reedling’s arch and match fantasy plants to sunny, damp and shaded places. Prepared saves are shortcuts for testing a later activity; record that they were used.

6. Try one useful adjustment: larger text, higher contrast, reduced motion, fewer philosophy invitations or local read aloud. Ask whether it helps. Do not require sound, narration or rapid pointing to finish an activity.

7. Ask the player what they enjoyed, what confused them, and what they would like to do next. Use Help → Playtest feedback to choose voluntary ratings and download the anonymous report. “Not rated” is a valid answer.

Observe separately from game points

• Enjoyment: desire to keep exploring, favourite activity, whether rewards felt earned and whether repetition became dull.
• Reading: words needing explanation; whether the situation and answer differences were understood; whether feedback clarified rather than judged.
• Learning: the player’s explanation of a consequence and a repair, including a reasonable alternative. A high game score alone is not evidence of real-world learning.
• Controls: help requested, missed buttons, unwanted movement, difficulty seeing a preview, camera trouble, and whether stopping a held button stops the player.
• Accessibility: text size, contrast, motion comfort, keyboard focus, ability to use non-drag controls, caption usefulness and read-aloud availability. Include players with varied access needs when possible; do not infer full accessibility from one session.
• Fairness: check that both respectful boundary options feel valid, that asking for help is welcomed, and that a wrong answer offers repair without shame or losing the world.

Device and input matrix

Record the actual device/browser tested; never fill an untested row as passed.

• Desktop keyboard and mouse: startup, camera, movement, ground build, preview-button build, every menu, save/reload, backup/restore.
• Desktop keyboard only: Tab and Shift+Tab, Enter and Space activation, on-screen one-step buttons, preview arrows, Place here, Escape, visible focus and sensible focus return after dialogs.
• iPhone/iPad Safari on real hardware: portrait and landscape, movement hold/release, two-finger diagonal movement, camera buttons, tap-to-preview + Place here, scrolling dialogs, device rotation, interruption/app switch, audio and saves.
• Android Chrome on real hardware: repeat the same checks; inspect performance and device heat during a longer session.
• Narrow-window developer inspection: useful for layout, but not evidence that iOS or Android touch events, sound or storage work.
• Assistive technology: test with intended keyboard/switch/screen-reader users. The 3D world has text directions and non-drag controls, but complete screen-reader play is not certified.
• Low-load graphics / reduced motion: ensure essential information still appears with sound off and decorative animation reduced.

Content review with an educator and parent

Review every prompt and explanation in each age profile. Check reading complexity, cultural assumptions, disability inclusion, privacy, boundaries, asking for help and competing reasonable choices. The fantasy garden’s sun/damp/shade rules are fictional game rules, not a botany lesson. Identify ambiguous answers before adding more content. Ask whether Wisdom scores encourage understanding or merely guessing.

Start with a few supported sessions across different ages, revise recurring problems, then repeat. Do not claim a representative study or educational effectiveness from a small convenience sample. An effectiveness study needs its own appropriate design and review.

Release decision

Keep the build in playtesting while there is save loss, a blocked essential activity, uncontrolled movement, inaccessible controls for the intended player group, distressing or misleading feedback, or serious device performance trouble. Fix recurring problems and retest them. Record remaining minor issues openly. A public release decision follows real device and participant evidence, not only passing automated tests.

Current verification boundaries

See Playtest-Build-QA.md for the exact developer checks and pending work. Physical mobile devices, actual participant sessions, audible voice quality and a formal accessibility assessment require human testing. No results for those are implied by this package.