How Wooden Sword Games Tested Mystic Duel Without an SDK
Wooden Sword Games runs BugHunter by Nodemori on Mystic Duel: Card Battler CCG, straight from the public store build, with no SDK and no changes to the game. The agent completed the tutorial, won a full battle and mapped 23 screen states.
Testing returned:
- Recorded sessions. Every free-play session came back with its video and console logs.
- Test cases. Plain-English steps with expected results, checked on the live game.
- Visual findings. In three free-play sessions run by the Nodemori team, an automatic re-check kept 7 of 12 flags, 6 distinct issues. No human verdict is recorded yet.
- Regression replays. Five of five passed with zero AI calls. Each was one or two steps long.
- Studio
- Mobile, registered in London
- Game
- 1v1 merge card battler
- Platform
- Mobile: iOS and Android
- Integration
- None

0
changes to the build: straight from the store
23
screen states mapped with no instructions
7 / 12
flags kept by an automatic re-check in Nodemori's sessions: 6 distinct issues
5 / 5
short replays passed: one or two steps each, zero AI calls
Why Merge Play Trips Up Scripted Bots
Wooden Sword Games is a mobile game studio registered in London. Its games include Puzzle Wars: Heroes, a match-3 RPG with competitive PvP, and smaller hybrid-casual titles such as Snack Bag and Snail Wars.
In Mystic Duel, three cards sit face up in the centre row and each player has four slots. Same-colour cards of the same rank merge into stronger heroes. Placing a card triggers an automatic attack, and abilities unlock from merge rank 2.
For a player, that is the fun. For a scripted bot, it is a row of traps: drags instead of taps, colour and rank rules on every move, a turn to read, and a lives gate that locks battles for hours after a loss. Beyond combat lie hero upgrades, leagues, Trophy Road, chests and a shop with secure store payment screens. The agent learned merge play from a short action specification, not a script.
Dealt In Early
BugHunter was playing Mystic Duel before there was a contract. It ran its first test cases, won a full battle, and then a scout run cleared the whole tutorial.

The team settled on the public build, and BugHunter installed it straight from the store. The first package followed: two gameplay videos, a generated design document and two specifications.
With that package in hand, the studio described what it wanted next. Watching thousands of playtest videos was not an option.
- Brief an agent the way it briefs a tester.Delivered
Plain-English test cases for a battle with Settings and exit, a purchase flow and opening the Shop.
- Play chosen situations to check balance.Still a goal
- Log wins and losses by league.Still a goal
- See the game on different screens without owning a hundred devices.Still a goal
One of the four requests is delivered. There is no record yet of batch balance runs or league logs, and in Wooden Sword's workspace Mystic Duel has run on one build and one device configuration.
Then came a dedicated environment with Mystic Duel installed and its specification preloaded, a live onboarding session with the studio's team, and a test-payment account so purchase flows could be tested without real charges. Today Wooden Sword runs BugHunter on a paid subscription.
“I needed to see it work without integration. BugHunter installed Mystic Duel straight from the store, with no SDK or special build. Before any contract, it was already merging cards and had won a full battle. That made me take it seriously.”
Four Ways the Agent Works the Board
01Let it explore

02Let it play

03Write the test like a tester's brief
Each step has an expected result that the agent checks on the live game. When it could not find a button, it stopped and named the missing element instead of guessing: “The settings button is not present on the current screen.”
- Start a battle
- Open Settings
- Exit
- Expect You Lose

04Record once, replay on demand

What Three Free-Play Sessions Found
Free play is not a demo mode. While the agent fights battles and walks the menus, it watches the screen for anything a player would notice. A second, automatic review then re-checks each flag and keeps it as a bug, marks it unclear or rejects it as a false positive.
The sample is three free-play sessions on the public 1.5.5 build, run by the Nodemori team on its own account over one week. They did not run in Wooden Sword's workspace. The agent raised 12 flags. The re-check kept 7 as bugs (58%), marked 4 unclear and rejected 1 as a false positive.
Kept means accepted by that automatic re-check. It is not a developer confirmation: no tester or developer has recorded a verdict on these flags, and no triage by the game's developers is recorded. Each kept flag has a screenshot and a video clip, so a tester can check it against the recording.
12
flags in 3 free-play sessions over one week
58%
kept as bugs by the automatic re-check: 7 of 12 flags, 6 distinct issues
7 / 7
kept flags came with a screenshot and a video clip
4
text issues
1
layout issue, flagged at two consecutive steps
1
render flag from a single frame
5
flags not kept: 4 unclear, 1 false positive
Two of the 7 flags are the same header overlap, captured at two consecutive steps, so the 7 flags describe 6 distinct issues. The agent rated all 7 not critical. The render flag rests on a single frame that may show an animation in progress, so it is counted but not shown below.
Two Issues, Step by Step
Two of the 6 issues, with the path the agent took to reach each one, from its step trace. The screenshots are crops of the ones the agent saved with each flag, with outlines and blurring added.

Grammar Errors in a Tutorial Line
Main screen, at the tutorial step that points to Heroes.
Steps the agent took
- On the main screen, claim the Stone League reward.
- Open the chest with Open All.
- Claim the new hero, Kieran the Assassin.
- Close the Get More Gems offer.
- Read the tutorial line that appears as the tutorial points to Heroes.
- Expected
- The line reads “Now you can equip an item on your hero.”
- Actual
- The line reads “Now you can equip item on you hero”: the article is missing and “your” is written as “you”.
Status: Kept as a bug by the automatic re-check and rated not critical by the agent. No human verdict and no studio triage are recorded.

Hero Title Drawn Over the Top Bar
Heroes, the hero panel of Bastian the Revenant, during the equipment tutorial.
Steps the agent took
- In the equipment tutorial, open Heroes, open a hero card and tap an equipment slot.
- Tap Shop and buy Heavy Armor Lvl.8.
- Go back to Heroes and open Bastian the Revenant.
- Tap the shield slot to open Inventory.
- Look at the top bar.
- Expected
- The top bar stays readable, with the player name and the gold counter visible. The hero title appears only inside the hero panel.
- Actual
- “Bastian the Revenant” is drawn over the player name and “Level 10” over the gold counter. The same overlap was captured at two consecutive steps and logged as two flags.
Status: Kept as a bug by the automatic re-check (two flags, one issue) and rated not critical by the agent. No human verdict and no studio triage are recorded.
The other three text issues are a Trophy Road line, “Finish all path and you can get main prize!”, and two lines with backticks in place of apostrophes: “Let`s go to the shop” and “your hero`s stats”. All 5 flags that were not kept were about promo pop-ups that open after a reward or after closing another offer.
What the Agent Brought Back
Across BugHunter's runs on Mystic Duel:
Zero integration work.
The public store build, as players get it: no SDK, no hooks, no special build.
Playing before the contract was signed.
Test cases, a won battle and the whole tutorial.
A map of the game.
23 screen states and 29 recorded transitions, from the tutorial to the shop. The specification lists seven game modes and six resources.
Seven flags kept by an automatic re-check.
Three free-play sessions run by the Nodemori team raised 12 flags. The re-check kept 7, which describe 6 distinct issues, each with a screenshot and a video clip. None has a human verdict yet.
Five of five short replays passed.
Each one or two steps long, 30 to 41 seconds, with zero AI calls. Three ran in Wooden Sword's own workspace.
Building a card battler or a merge game?
Start with a free pilot on your own game.