[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7E6FSW7ZYP2CAMYWJE4NQ2W":3,"skill-same:01M7E6FSW7ZYP2CAMYWJE4NQ2W":47},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":29,"ownerHtmlUrl":30,"ownerBlogUrl":31,"ownerDescription":6,"updatedAt":32,"install":33,"resources":46},"01M7E6FSW7ZYP2CAMYWJE4NQ2W","automation-hooks",null,"Use when a Godot game needs command-line flags, a bot that plays it, a recorded solution replayed in CI, or balance and A\u002FB results from many runs, so it can be played and measured without a person.","nick-vas\u002Fhyperdot","nick-vas","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F7181961?v=4","01M7E6F6F0BQ49V0JZHX9BXCBD","Nikolaos Vasilikiotis","01M7E6F6FZW8SW1AJ1S7ARFE89",1,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fnick-vas\u002Fhyperdot\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fautomation-hooks\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: automation-hooks\ndescription: Use when a Godot game needs command-line flags, a bot that plays it, a recorded solution replayed in CI, or balance and A\u002FB results from many runs, so it can be played and measured without a person.\n---\n\n# Automation hooks: flags, seeds, bots, logs\n\nAn agent can only check what the game lets it drive. A handful of command-line flags turn the game itself into the test harness:\nheadless balance runs, reproducible bugs, benchmarks and screenshots. Build them early (the vertical slice gate asks for them).\n\n## Flags go after `--`\n\nGodot ignores unknown engine flags silently, and **release exports ignore scene paths** on the command line. So every debug entry\npoint is a user flag after `--`, read with `OS.GetCmdlineUserArgs()`:\n\n```csharp\npublic static class Flags\n{\n    static readonly string[] Args = OS.GetCmdlineUserArgs();\n    public static bool Has(string name) => Array.IndexOf(Args, $\"--{name}\") >= 0;\n    public static string Get(string name, string fallback = \"\")\n    {\n        foreach (var a in Args) if (a.StartsWith($\"--{name}=\")) return a[(name.Length + 3)..];\n        return fallback;\n    }\n}\n\u002F\u002F godot --path . -- --autoplay --seed=42 --act=3 --no-save\n```\n\nThe kit passes the same flags everywhere with `-GameArgs` (comma-separated, so one value cannot contain a comma):\n\n```powershell\npwsh ${CLAUDE_PLUGIN_ROOT}\u002Fci\u002Fgodot\u002Fhyperdot.ps1 test -Json -Quiet -Runners smoke -GameArgs '--autoplay','--seed=42','--no-save'\npwsh ${CLAUDE_PLUGIN_ROOT}\u002Fci\u002Fgodot\u002Fhyperdot.ps1 test -Json -Quiet -Runners playthrough -GameArgs '--bot-skill=0.5'\npwsh ${CLAUDE_PLUGIN_ROOT}\u002Fci\u002Fgodot\u002Fhyperdot.ps1 bench -Json -Quiet -GameArgs '--stress'\npwsh ${CLAUDE_PLUGIN_ROOT}\u002Fci\u002Fgodot\u002Fhyperdot.ps1 shot -Json -Quiet -Scene res:\u002F\u002Fui\u002FMap.tscn -GameArgs '--seed=7'\n```\n\nThese flags are debug tools, read only under `#if DEBUG` (`debug-tools`). `test` (its `smoke` and `playthrough`\nrunners), `bench` and `shot` run the project with the editor binary, a debug build, so the flags work there. `export -Run`\nlaunches the release export: it proves that build starts and runs without errors, and a bot flag passed to it does\nnothing.\n\n## Flags worth having\n\n| Flag | Why |\n|---|---|\n| `--seed=N` | reproduces a whole run; every bug report and balance run carries one |\n| `--autoplay`, `--bot-skill=0.5` | a bot plays, so smoke runs and exports exercise real gameplay |\n| `--no-save` | test runs never touch the player's save |\n| `--act=N`, `--start-floor=N`, `--level=X` | jump straight to the content under test |\n| `--stress` | the worst case for `bench` (`performance`) |\n| `--capture=nohud` | hides everything in group `hud` (put the minimap and waypoint markers there too): navigation playtests run with it (`game-design` section 7); capture mode is `debug-tools` section 7 |\n| `--shot=path --shot-at=seconds` | the game saves its own screenshot (inside a real run, with game state); `shot` covers single scenes |\n| `--playback=\u003Cfile>`, `--skip-to-frame=N` | replays an `InputLog` the game saved on quit or on an error (`determinism`); the skip plays N ticks undrawn, then hands control back, still recording |\n\nEvery session records its input, so a bug found by hand ends as a file: quit, then `-- --playback=user:\u002F\u002Freplays\u002Flast.replay\n--skip-to-frame=N` lands on the frame of the bug. Playback also means `--no-save` and the C# example's `LocalPlatform` in place of Steam,\nnever stored, so a replay unlocks and posts nothing real. Unlike `--replay` below, which stores intent and survives level edits,\nan input log holds raw ticks: it needs the same simulation rules (or its `Rules` branch) to play.\n\n## Deterministic randomness\n\nOne seeded `RandomNumberGenerator` per decision point, derived from the run seed and the point's identity, so changing one level\ndoes not reshuffle all the others: `rng.Seed = Seeds.Derive(runSeed, act, floor, column);` (`procgen` has the\nten-line `Seeds` class). Never `HashCode.Combine` or `GetHashCode()` for this: .NET randomises them per process, so the same\nseed gives a different run on every launch. Never use the global `GD.Randf()` for gameplay.\n\n## Bots for balance\n\n- A perfect bot is useless for balance. Add human-like error: a reaction delay (0.2-0.45 s) and a misjudged landing, both scaled\n  by `--bot-skill`. Around 0.5, it should lose the way a decent player does.\n- Log **game time**, not wall-clock, and run many seeds in parallel faster than real time:\n  `godot --headless --path . --fixed-fps 60 -- --autoplay --seed=N` (an engine flag, so it goes before `--`); the kit's\n  `smoke -FixedFps 60` does the same.\n- **Prove the bot finishes.** The bot prints a line per milestone (`[Run] act 1 done`, `[Run] ending reached`) and\n  `playthrough.json` lists them; the `playthrough` runner fails when one is missing, which `smoke` does not notice. It also\n  meets the plan gate `production.playthrough`.\n- Deterministic bots loop: a \"safe\" fallback move repeated forever. When the goal is out of reach, pick a *random* safe move.\n- Bots press the player's inputs (the same actions or input frames), never set state: they can do nothing a player\n  cannot, and what they find is a player's bug. For Honor's bots used the player's input pipeline for this reason.\n- **Bot against bot for difficulty tiers.** For Honor ran over 250 bot duels a night on one PC and compared each\n  matchup's win ratio with a target (a fast fighter should beat a slow one more often than not), not with 50%. Anchor\n  each difficulty with one bot tuned by hand against people: two broken bots agree with each other. It finds large\n  anomalies, not fine balance.\n- Test every level or encounter end to end with the bot at two skill levels and several seeds. In one project this found five\n  bugs that unit tests and solvers never showed.\n- The bot presses InputMap actions, never calls the player's methods, so the player code is what gets tested and edits to\n  it do not break the bot (Battlefield V's bots, Jonas Gillberg, EA, GDC): `Input.ActionPress`\u002F`ActionRelease` for\n  code that polls `IsActionPressed`, `Input.ParseInputEvent` for code in `_UnhandledInput` (`input`). A bot that\n  never opens menus does not test them (UI crashes, 13% of a playtest's, were all missed by those bots); `flows` does.\n- Unstick, then report: no progress toward the next waypoint for a set time means jump or interact, then teleport,\n  and log `[Stuck] pos=x,y,z level=\u003Cname>`. Grepped over many runs, those lines map where players get trapped in\n  collision.\n- On spawn the bot picks a random loadout, weighted toward the default one, so it covers the combinations testers skip.\n- Before trusting a balance change, run the unchanged build on two separate seed sets (an A\u002FA run): their gap is the noise a\n  change must beat. SocialPoint's A\u002FA splits of real players differed by 5 to 10% (Aviv Stern, GDC). With few runs, judge\n  yes\u002Fno outcomes (cleared, won), which need far fewer runs than amounts (gold, seconds).\n\n## Comparing two versions\n\nDid the change help (a new jump, an easier puzzle, a shop coupon)? Two averages from a few runs differ by chance alone, so let a\ntest decide. `${CLAUDE_PLUGIN_ROOT}\u002Fexamples\u002Fcsharp\u002Fsrc\u002FExample.Core\u002FAnalytics\u002FStats.cs` has them, engine-free, with\n`StatsTests`; pick by what is measured:\n\n| Outcome | Test |\n|---|---|\n| a number with a real mean (seconds to clear, damage taken) | `Stats.WelchTTest`; `Stats.PairedTTest` when the same seeds or players saw both versions |\n| yes or no (cleared, bought, quit) | `Stats.RelativeRisk` (\"2.4 times as likely\", to convince others), `Stats.ProportionDifference` (points, to decide) |\n| skewed (a few marathon players), a 1 to 5 rating, a ranking | `Stats.MannWhitney`; for ratings, do not pick the test after seeing the data |\n\n- Fix the metric, the threshold (5%) and the number of runs first, then read the result once. Stopping the first time p\n  drops under 5% gave 19% false positives over ten looks (`StatsTests`). Several variants share that 5%: divide it.\n- Every p-value is two-tailed: a change can make things worse too.\n- Read the interval, not only p. When it straddles zero (one for a ratio), `IsResult` is false: no result, whatever the\n  mean says. A real difference can still be too small to be worth its cost.\n- Bots: run both versions on the same seeds and pair them. On the same data, pairing turned no result into p = 0.003.\n- One question per test: twenty metrics at 5% each find a \"difference\" by chance. Balancing many PvP classes is beyond this.\n- The tests assume independent samples with final outcomes: players in one multiplayer match affect each other (compare per\n  match), and an outcome that can still happen (a purchase on day 3) is read only once its running total flattens (Swrve's\n  A\u002FB testing talk, GDC).\n- With real players (Emily Greer, Kongregate, GDC): assign the variant when a player first meets the feature (opens the\n  shop), not at launch; run long enough for less engaged players to arrive, since the first days bring the best ones; read\n  every downstream metric, not only the target (a simpler shop raised buyers 9% and cut revenue 11%; a flash sale read +8%\n  revenue at 10 days and -2% at 30); judge spend with medians and `Stats.MannWhitney`, as one big buyer moves a mean.\n\n## Reading balance from run data\n\nPick and win rates per card, relic or upgrade, from one summary per finished run (bots or players): `BalanceReport` in\n`${CLAUDE_PLUGIN_ROOT}\u002Fexamples\u002Fcsharp\u002Fsrc\u002FExample.Core\u002FAnalytics\u002F`, with `BalanceReportTests`. Plain averages lie four ways:\n\n- **Late options flatter themselves.** An option found near the end is seen mostly in runs that were already winning (Slay the\n  Spire's Madness, from an event just before the last boss). Compare its runs only with all runs that reached the floor where\n  it was taken (`FloorMatchedLift`): a do-nothing option offered on floor 8 of 10 won 74% against 36% overall, lift 0.00.\n- **Heavy players drown out everyone.** Two testers with 500 runs each made a 23% win rate (per player) read 84% (per run).\n  Weigh each player, or each session when there is no player id, once (`PerSession`).\n- **Difficulties differ.** Report each difficulty apart; pooled, a neutral option favoured on the hard one looked 16 points worse.\n- **The audience shifts.** A total moves more often because who plays changed (where players came from, platform, days\n  since install) than because the game did (Greer). Split by those, show each group's sample size, and ignore groups too\n  small to mean anything.\n- The goal is not equal power: every option should be worth taking some of the time, so look for pick rates near 0 or 1 when\n  offered. Rare, hard-to-assemble combos may be strong. Metrics do not show what is boring (slow, grindy, optimal loops): watch\n  play for that. Decide whether a change helped with the tests above, not by eye.\n\n## What studios test\n\n- **The whole game, every build.** Croteam ran a bot that solved all of The Talos Principle on every build; Dragon Quest III\n  nearly shipped with nobody having played it to the end. Keep one flow or bot run that reaches the ending in CI.\n- **Balance in sequence.** Dragon Quest II's fights were each fair alone and too hard back to back: balance runs chain the\n  encounters of a stage or a whole run, not one fight.\n- **Systems, not paths.** Interlinked systems (Animal Crossing's calendar, an A-Life world) are not covered by playing in\n  order: add seeded random-input and long soak runs, and jump the clock.\n- **Why players fail.** Log every death or quit with its cause (`[Death] cause=fall level=3 t=212.4`); a heat map of deaths\n  and movement shows where the design does not work as planned (GSC on S.T.A.L.K.E.R. 2, DayZ).\n- **Questions, not personas.** Horizon Zero Dawn's testers playing a \"hoarder\" for four or five days found what a\n  four-hour inventory stress task found. Give each exploratory session one question to answer (the charter), a time box,\n  the results, and follow-up questions as new sessions (Guerrilla's QA manager, GDC).\n- **Which build.** Stamp the commit and build time on screen and in every log, so a report or screenshot names its build.\n- **A feedback key.** One key and one typed line (\"these rocks look wrong\") save a screenshot, the log, the save file and\n  a `user:\u002F\u002Freports\u002F\u003Ctime>.json` with the scene path, the player's and camera's `GlobalTransform`, the seed and the\n  commit (Inkbound, Croteam). The place and view are what make a vague report fixable; the fixing commit names the report.\n  Copy `${CLAUDE_PLUGIN_ROOT}\u002Fci\u002Fgodot\u002Ftemplates\u002Faddons\u002Fhyperdot_report_pins\u002F` to `res:\u002F\u002Faddons\u002F`: its `bug_report.gd`\n  saves that report (2D or 3D), and the editor add-on pins each report on its scene, with a Reports dock. Share a\n  report in `res:\u002F\u002Freports\u002F` with an empty `.gdignore`, so it is never exported.\n\n## Record a solution, replay it in CI\n\nCroteam's bot finished The Talos Principle in about 20 minutes on every build (a person needs about ten hours), so a jump\nor physics change was known safe within the hour. Its recordings survived level edits because they stored intent:\n\n- `--record=res:\u002F\u002Fsolutions\u002F\u003Clevel>.json` writes the designer's solution as steps on named nodes (`take Key_A`,\n  `place Box_1 on Socket_2`, `use Door_3`), never raw input or coordinates.\n- `--replay=\u003Cfile>` walks to each target with `NavigationAgent3D`, performs the step, and logs `[Run] level=\u003Cname> solved`.\n  By hand, run it under `--fixed-fps 60` (fast-forward with fixed physics steps).\n- A `smoke` run passes on a bot stuck in the first room; the `playthrough` runner does not. Put the replay in\n  `playthrough.json` (`\"args\": [\"--replay=all\", \"--no-save\"]`, `\"expect\"` with each level's `[Run] level=\u003Cname> solved`\n  and the ending last, `\"frames\"` sized from a local run): it fails when a milestone is missing (Prove the bot finishes,\n  above).\n- Let small content changes through (Final Fantasy VII Remake's replays, Square Enix, GDC): a step marked `optional` (a\n  first-time tutorial) is dropped when it never comes, an unexpected scene the game can skip (a new cutscene or fight) is\n  skipped instead of failing, and a step times out at a multiple of its recorded duration, not a fixed limit. When the\n  game is not where the recording expects, try the other legal steps for a while before failing, and log\n  `[Run] step=N repaired` (NetEase's quest bot fell back to searching that way).\n- Fail a level whose replay runs far past its budget even without an error: it would stall a player too (NetEase flagged\n  any quest not done in 15 minutes).\n- It proves the game can be finished, not that it has no bugs (their door stopped opening after a texture change).\n  People still play it. It proves one route: NetEase's shortest path never jumped down the cliff that players could not\n  climb back up, so add the seeded random bot. It proves nothing about pictures: Square Enix's passed on a black\n  screen, so save `--shot` pictures at named steps and look at them.\n- A hung run (a deadlock, an `await` that never resumes) is stopped after `HYPERDOT_TIMEOUT` seconds (default 1800) and\n  fails; set it a little above the run's real length so a hang is caught early.\n\n## One log line per meaningful event\n\nTagged and grep-able: `[Map] node=3 kind=elite`, `[Fight] won in 142.5s hp=31`, `[Run] act 2 cleared in 1410s`. Balance and\nregression checks become a `grep` over smoke and bot logs (they are in `test-results\u002Fsmoke\u002Fsmoke.log`).\n\n## Checklist\n\n- [ ] `--seed`, `--autoplay`, `--no-save` exist and are read with `OS.GetCmdlineUserArgs()`\n- [ ] `test -Json -GameArgs '--autoplay','--no-save'` passes (the bot played without errors)\n- [ ] a stress flag exists for `bench`\n- [ ] events are logged with tags\n","# Automation hooks: flags, seeds, bots, logs\n\nAn agent can only check what the game lets it drive. A handful of command-line flags turn the game itself into the test harness:\nheadless balance runs, reproducible bugs, benchmarks and screenshots. Build them early (the vertical slice gate asks for them).\n\n## Flags go after `--`\n\nGodot ignores unknown engine flags silently, and **release exports ignore scene paths** on the command line. So every debug entry\npoint is a user flag after `--`, read with `OS.GetCmdlineUserArgs()`:\n\n```csharp\npublic static class Flags\n{\n    static readonly string[] Args = OS.GetCmdlineUserArgs();\n    public static bool Has(string name) => Array.IndexOf(Args, $\"--{name}\") >= 0;\n    public static string Get(string name, string fallback = \"\")\n    {\n        foreach (var a in Args) if (a.StartsWith($\"--{name}=\")) return a[(name.Length + 3)..];\n        return fallback;\n    }\n}\n\u002F\u002F godot --path . -- --autoplay --seed=42 --act=3 --no-save\n```\n\nThe kit passes the same flags everywhere with `-GameArgs` (comma-separated, so one value cannot contain a comma):\n\n```powershell\npwsh ${CLAUDE_PLUGIN_ROOT}\u002Fci\u002Fgodot\u002Fhyperdot.ps1 test -Json -Quiet -Runners smoke -GameArgs '--autoplay','--seed=42','--no-save'\npwsh ${CLAUDE_PLUGIN_ROOT}\u002Fci\u002Fgodot\u002Fhyperdot.ps1 test -Json -Quiet -Runners playthrough -GameArgs '--bot-skill=0.5'\npwsh ${CLAUDE_PLUGIN_ROOT}\u002Fci\u002Fgodot\u002Fhyperdot.ps1 bench -Json -Quiet -GameArgs '--stress'\npwsh ${CLAUDE_PLUGIN_ROOT}\u002Fci\u002Fgodot\u002Fhyperdot.ps1 shot -Json -Quiet -Scene res:\u002F\u002Fui\u002FMap.tscn -GameArgs '--seed=7'\n```\n\nThese flags are debug tools, read only under `#if DEBUG` (`debug-tools`). `test` (its `smoke` and `playthrough`\nrunners), `bench` and `shot` run the project with the editor binary, a debug build, so the flags work there. `export -Run`\nlaunches the release export: it proves that build starts and runs without errors, and a bot flag passed to it does\nnothing.\n\n## Flags worth having\n\n| Flag | Why |\n|---|---|\n| `--seed=N` | reproduces a whole run; every bug report and balance run carries one |\n| `--autoplay`, `--bot-skill=0.5` | a bot plays, so smoke runs and exports exercise real gameplay |\n| `--no-save` | test runs never touch the player's save |\n| `--act=N`, `--start-floor=N`, `--level=X` | jump straight to the content under test |\n| `--stress` | the worst case for `bench` (`performance`) |\n| `--capture=nohud` | hides everything in group `hud` (put the minimap and waypoint markers there too): navigation playtests run with it (`game-design` section 7); capture mode is `debug-tools` section 7 |\n| `--shot=path --shot-at=seconds` | the game saves its own screenshot (inside a real run, with game state); `shot` covers single scenes |\n| `--playback=\u003Cfile>`, `--skip-to-frame=N` | replays an `InputLog` the game saved on quit or on an error (`determinism`); the skip plays N ticks undrawn, then hands control back, still recording |\n\nEvery session records its input, so a bug found by hand ends as a file: quit, then `-- --playback=user:\u002F\u002Freplays\u002Flast.replay\n--skip-to-frame=N` lands on the frame of the bug. Playback also means `--no-save` and the C# example's `LocalPlatform` in place of Steam,\nnever stored, so a replay unlocks and posts nothing real. Unlike `--replay` below, which stores intent and survives level edits,\nan input log holds raw ticks: it needs the same simulation rules (or its `Rules` branch) to play.\n\n## Deterministic randomness\n\nOne seeded `RandomNumberGenerator` per decision point, derived from the run seed and the point's identity, so changing one level\ndoes not reshuffle all the others: `rng.Seed = Seeds.Derive(runSeed, act, floor, column);` (`procgen` has the\nten-line `Seeds` class). Never `HashCode.Combine` or `GetHashCode()` for this: .NET randomises them per process, so the same\nseed gives a different run on every launch. Never use the global `GD.Randf()` for gameplay.\n\n## Bots for balance\n\n- A perfect bot is useless for balance. Add human-like error: a reaction delay (0.2-0.45 s) and a misjudged landing, both scaled\n  by `--bot-skill`. Around 0.5, it should lose the way a decent player does.\n- Log **game time**, not wall-clock, and run many seeds in parallel faster than real time:\n  `godot --headless --path . --fixed-fps 60 -- --autoplay --seed=N` (an engine flag, so it goes before `--`); the kit's\n  `smoke -FixedFps 60` does the same.\n- **Prove the bot finishes.** The bot prints a line per milestone (`[Run] act 1 done`, `[Run] ending reached`) and\n  `playthrough.json` lists them; the `playthrough` runner fails when one is missing, which `smoke` does not notice. It also\n  meets the plan gate `production.playthrough`.\n- Deterministic bots loop: a \"safe\" fallback move repeated forever. When the goal is out of reach, pick a *random* safe move.\n- Bots press the player's inputs (the same actions or input frames), never set state: they can do nothing a player\n  cannot, and what they find is a player's bug. For Honor's bots used the player's input pipeline for this reason.\n- **Bot against bot for difficulty tiers.** For Honor ran over 250 bot duels a night on one PC and compared each\n  matchup's win ratio with a target (a fast fighter should beat a slow one more often than not), not with 50%. Anchor\n  each difficulty with one bot tuned by hand against people: two broken bots agree with each other. It finds large\n  anomalies, not fine balance.\n- Test every level or encounter end to end with the bot at two skill levels and several seeds. In one project this found five\n  bugs that unit tests and solvers never showed.\n- The bot presses InputMap actions, never calls the player's methods, so the player code is what gets tested and edits to\n  it do not break the bot (Battlefield V's bots, Jonas Gillberg, EA, GDC): `Input.ActionPress`\u002F`ActionRelease` for\n  code that polls `IsActionPressed`, `Input.ParseInputEvent` for code in `_UnhandledInput` (`input`). A bot that\n  never opens menus does not test them (UI crashes, 13% of a playtest's, were all missed by those bots); `flows` does.\n- Unstick, then report: no progress toward the next waypoint for a set time means jump or interact, then teleport,\n  and log `[Stuck] pos=x,y,z level=\u003Cname>`. Grepped over many runs, those lines map where players get trapped in\n  collision.\n- On spawn the bot picks a random loadout, weighted toward the default one, so it covers the combinations testers skip.\n- Before trusting a balance change, run the unchanged build on two separate seed sets (an A\u002FA run): their gap is the noise a\n  change must beat. SocialPoint's A\u002FA splits of real players differed by 5 to 10% (Aviv Stern, GDC). With few runs, judge\n  yes\u002Fno outcomes (cleared, won), which need far fewer runs than amounts (gold, seconds).\n\n## Comparing two versions\n\nDid the change help (a new jump, an easier puzzle, a shop coupon)? Two averages from a few runs differ by chance alone, so let a\ntest decide. `${CLAUDE_PLUGIN_ROOT}\u002Fexamples\u002Fcsharp\u002Fsrc\u002FExample.Core\u002FAnalytics\u002FStats.cs` has them, engine-free, with\n`StatsTests`; pick by what is measured:\n\n| Outcome | Test |\n|---|---|\n| a number with a real mean (seconds to clear, damage taken) | `Stats.WelchTTest`; `Stats.PairedTTest` when the same seeds or players saw both versions |\n| yes or no (cleared, bought, quit) | `Stats.RelativeRisk` (\"2.4 times as likely\", to convince others), `Stats.ProportionDifference` (points, to decide) |\n| skewed (a few marathon players), a 1 to 5 rating, a ranking | `Stats.MannWhitney`; for ratings, do not pick the test after seeing the data |\n\n- Fix the metric, the threshold (5%) and the number of runs first, then read the result once. Stopping the first time p\n  drops under 5% gave 19% false positives over ten looks (`StatsTests`). Several variants share that 5%: divide it.\n- Every p-value is two-tailed: a change can make things worse too.\n- Read the interval, not only p. When it straddles zero (one for a ratio), `IsResult` is false: no result, whatever the\n  mean says. A real difference can still be too small to be worth its cost.\n- Bots: run both versions on the same seeds and pair them. On the same data, pairing turned no result into p = 0.003.\n- One question per test: twenty metrics at 5% each find a \"difference\" by chance. Balancing many PvP classes is beyond this.\n- The tests assume independent samples with final outcomes: players in one multiplayer match affect each other (compare per\n  match), and an outcome that can still happen (a purchase on day 3) is read only once its running total flattens (Swrve's\n  A\u002FB testing talk, GDC).\n- With real players (Emily Greer, Kongregate, GDC): assign the variant when a player first meets the feature (opens the\n  shop), not at launch; run long enough for less engaged players to arrive, since the first days bring the best ones; read\n  every downstream metric, not only the target (a simpler shop raised buyers 9% and cut revenue 11%; a flash sale read +8%\n  revenue at 10 days and -2% at 30); judge spend with medians and `Stats.MannWhitney`, as one big buyer moves a mean.\n\n## Reading balance from run data\n\nPick and win rates per card, relic or upgrade, from one summary per finished run (bots or players): `BalanceReport` in\n`${CLAUDE_PLUGIN_ROOT}\u002Fexamples\u002Fcsharp\u002Fsrc\u002FExample.Core\u002FAnalytics\u002F`, with `BalanceReportTests`. Plain averages lie four ways:\n\n- **Late options flatter themselves.** An option found near the end is seen mostly in runs that were already winning (Slay the\n  Spire's Madness, from an event just before the last boss). Compare its runs only with all runs that reached the floor where\n  it was taken (`FloorMatchedLift`): a do-nothing option offered on floor 8 of 10 won 74% against 36% overall, lift 0.00.\n- **Heavy players drown out everyone.** Two testers with 500 runs each made a 23% win rate (per player) read 84% (per run).\n  Weigh each player, or each session when there is no player id, once (`PerSession`).\n- **Difficulties differ.** Report each difficulty apart; pooled, a neutral option favoured on the hard one looked 16 points worse.\n- **The audience shifts.** A total moves more often because who plays changed (where players came from, platform, days\n  since install) than because the game did (Greer). Split by those, show each group's sample size, and ignore groups too\n  small to mean anything.\n- The goal is not equal power: every option should be worth taking some of the time, so look for pick rates near 0 or 1 when\n  offered. Rare, hard-to-assemble combos may be strong. Metrics do not show what is boring (slow, grindy, optimal loops): watch\n  play for that. Decide whether a change helped with the tests above, not by eye.\n\n## What studios test\n\n- **The whole game, every build.** Croteam ran a bot that solved all of The Talos Principle on every build; Dragon Quest III\n  nearly shipped with nobody having played it to the end. Keep one flow or bot run that reaches the ending in CI.\n- **Balance in sequence.** Dragon Quest II's fights were each fair alone and too hard back to back: balance runs chain the\n  encounters of a stage or a whole run, not one fight.\n- **Systems, not paths.** Interlinked systems (Animal Crossing's calendar, an A-Life world) are not covered by playing in\n  order: add seeded random-input and long soak runs, and jump the clock.\n- **Why players fail.** Log every death or quit with its cause (`[Death] cause=fall level=3 t=212.4`); a heat map of deaths\n  and movement shows where the design does not work as planned (GSC on S.T.A.L.K.E.R. 2, DayZ).\n- **Questions, not personas.** Horizon Zero Dawn's testers playing a \"hoarder\" for four or five days found what a\n  four-hour inventory stress task found. Give each exploratory session one question to answer (the charter), a time box,\n  the results, and follow-up questions as new sessions (Guerrilla's QA manager, GDC).\n- **Which build.** Stamp the commit and build time on screen and in every log, so a report or screenshot names its build.\n- **A feedback key.** One key and one typed line (\"these rocks look wrong\") save a screenshot, the log, the save file and\n  a `user:\u002F\u002Freports\u002F\u003Ctime>.json` with the scene path, the player's and camera's `GlobalTransform`, the seed and the\n  commit (Inkbound, Croteam). The place and view are what make a vague report fixable; the fixing commit names the report.\n  Copy `${CLAUDE_PLUGIN_ROOT}\u002Fci\u002Fgodot\u002Ftemplates\u002Faddons\u002Fhyperdot_report_pins\u002F` to `res:\u002F\u002Faddons\u002F`: its `bug_report.gd`\n  saves that report (2D or 3D), and the editor add-on pins each report on its scene, with a Reports dock. Share a\n  report in `res:\u002F\u002Freports\u002F` with an empty `.gdignore`, so it is never exported.\n\n## Record a solution, replay it in CI\n\nCroteam's bot finished The Talos Principle in about 20 minutes on every build (a person needs about ten hours), so a jump\nor physics change was known safe within the hour. Its recordings survived level edits because they stored intent:\n\n- `--record=res:\u002F\u002Fsolutions\u002F\u003Clevel>.json` writes the designer's solution as steps on named nodes (`take Key_A`,\n  `place Box_1 on Socket_2`, `use Door_3`), never raw input or coordinates.\n- `--replay=\u003Cfile>` walks to each target with `NavigationAgent3D`, performs the step, and logs `[Run] level=\u003Cname> solved`.\n  By hand, run it under `--fixed-fps 60` (fast-forward with fixed physics steps).\n- A `smoke` run passes on a bot stuck in the first room; the `playthrough` runner does not. Put the replay in\n  `playthrough.json` (`\"args\": [\"--replay=all\", \"--no-save\"]`, `\"expect\"` with each level's `[Run] level=\u003Cname> solved`\n  and the ending last, `\"frames\"` sized from a local run): it fails when a milestone is missing (Prove the bot finishes,\n  above).\n- Let small content changes through (Final Fantasy VII Remake's replays, Square Enix, GDC): a step marked `optional` (a\n  first-time tutorial) is dropped when it never comes, an unexpected scene the game can skip (a new cutscene or fight) is\n  skipped instead of failing, and a step times out at a multiple of its recorded duration, not a fixed limit. When the\n  game is not where the recording expects, try the other legal steps for a while before failing, and log\n  `[Run] step=N repaired` (NetEase's quest bot fell back to searching that way).\n- Fail a level whose replay runs far past its budget even without an error: it would stall a player too (NetEase flagged\n  any quest not done in 15 minutes).\n- It proves the game can be finished, not that it has no bugs (their door stopped opening after a texture change).\n  People still play it. It proves one route: NetEase's shortest path never jumped down the cliff that players could not\n  climb back up, so add the seeded random bot. It proves nothing about pictures: Square Enix's passed on a black\n  screen, so save `--shot` pictures at named steps and look at them.\n- A hung run (a deadlock, an `await` that never resumes) is stopped after `HYPERDOT_TIMEOUT` seconds (default 1800) and\n  fails; set it a little above the run's real length so a hang is caught early.\n\n## One log line per meaningful event\n\nTagged and grep-able: `[Map] node=3 kind=elite`, `[Fight] won in 142.5s hp=31`, `[Run] act 2 cleared in 1410s`. Balance and\nregression checks become a `grep` over smoke and bot logs (they are in `test-results\u002Fsmoke\u002Fsmoke.log`).\n\n## Checklist\n\n- [ ] `--seed`, `--autoplay`, `--no-save` exist and are read with `OS.GetCmdlineUserArgs()`\n- [ ] `test -Json -GameArgs '--autoplay','--no-save'` passes (the bot played without errors)\n- [ ] a stress flag exists for `bench`\n- [ ] events are logged with tags\n",{"name":5,"description":7},"main","5fa4c2262d814d50f72198a95b8c483ff195f90f",".agents\u002Fskills\u002Fautomation-hooks",0,"https:\u002F\u002Fgithub.com\u002Fnick-vas\u002Fhyperdot","https:\u002F\u002Fnick-vas.github.io\u002Fhyperdot\u002F","https:\u002F\u002Fgithub.com\u002Fnick-vas","https:\u002F\u002Fcommon-beef.itch.io\u002F","2026-11-13 18:55:29",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add nick-vas\u002Fhyperdot --skill automation-hooks",{"tool":39,"command":40},"bunx","bunx skills add nick-vas\u002Fhyperdot --skill automation-hooks",{"tool":42,"command":43},"pnpm","pnpm dlx skills add nick-vas\u002Fhyperdot --skill automation-hooks","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fnick-vas\u002Fhyperdot\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fautomation-hooks\u002FSKILL.md",true,[],[48,55,61,68,75,82],{"skillNo":49,"name":50,"nameZh":6,"description":51,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":52,"categories":53,"keywords":54,"publishedAt":19},"01M7E6RF2S03S92PB9RZB4S4W0","over-budget","Use when a fixture exceeds the size budget.","https:\u002F\u002Fgithub.com\u002Fnick-vas\u002Fhyperdot\u002Fblob\u002Fmain\u002Ftests\u002Fskill-fixtures\u002Fover-budget\u002FSKILL.md",[],[],{"skillNo":56,"name":57,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":58,"categories":59,"keywords":60,"publishedAt":19},"01M7E6RBY8SDNMA3ZARE8MAH0G","bad-yaml","https:\u002F\u002Fgithub.com\u002Fnick-vas\u002Fhyperdot\u002Fblob\u002Fmain\u002Ftests\u002Fskill-fixtures\u002Fbad-yaml\u002FSKILL.md",[],[],{"skillNo":62,"name":63,"nameZh":6,"description":64,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":65,"categories":66,"keywords":67,"publishedAt":19},"01M7E6R9S9YHTC552SR41RQAJ1","bad-workflow-description","Use when testing; runs the kit and then reads the JSON.","https:\u002F\u002Fgithub.com\u002Fnick-vas\u002Fhyperdot\u002Fblob\u002Fmain\u002Ftests\u002Fskill-fixtures\u002Fbad-workflow-description\u002FSKILL.md",[],[],{"skillNo":69,"name":70,"nameZh":6,"description":71,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":72,"categories":73,"keywords":74,"publishedAt":19},"01M7E6R6CWSAN48C80130CM1AK","bad-version","Use when a fixture quotes a stale package version.","https:\u002F\u002Fgithub.com\u002Fnick-vas\u002Fhyperdot\u002Fblob\u002Fmain\u002Ftests\u002Fskill-fixtures\u002Fbad-version\u002FSKILL.md",[],[],{"skillNo":76,"name":77,"nameZh":6,"description":78,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"publishedAt":19},"01M7E6R2C0E24TC133ESHKZT9R","bad-value","Use when a fixture passes a value outside a parameter's set.","https:\u002F\u002Fgithub.com\u002Fnick-vas\u002Fhyperdot\u002Fblob\u002Fmain\u002Ftests\u002Fskill-fixtures\u002Fbad-value\u002FSKILL.md",[],[],{"skillNo":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":88,"publishedAt":19},"01M7E6QYM81V061X7DSQ94CZRP","bad-trigger","Helps with Godot tests.","https:\u002F\u002Fgithub.com\u002Fnick-vas\u002Fhyperdot\u002Fblob\u002Fmain\u002Ftests\u002Fskill-fixtures\u002Fbad-trigger\u002FSKILL.md",[],[]]