FloatSoda: Skill for Claude Code

.agents/skills/floatsoda-junior-coder-test/SKILL.md

floatsoda-junior-coder-test is a skill for Claude Code, Codex from sumx21t-3310/FloatSoda. It costs 309 tokens per session (3,773 once invoked), scanned A, original, MIT.

A black-box test of FloatSoda's public documentation and API, using a less experienced coding model as a stand-in for a beginner. Black-box means the model can use only the public docs, not the library's source code.

In plain words
What is it for?
Use it to test public docs with a realistic VRChatter feature request, run the generated code in an isolated project, and investigate every failure without exposing the implementation to the test model.
Why use it?
It reveals whether a realistic user or coding assistant can build a feature without being misled by unclear documentation or an easy-to-misuse API. Failures are then checked against the real source to distinguish documentation problems, library bugs, and design issues.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one. Also seen: installed under .agents/ (shared by several agents).

This is sumx21t-3310/FloatSoda's own configuration. It tells Claude Code and Codex how to work on FloatSoda itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything FloatSoda configures →

Reuse

Borrowing it

Nothing to install: this file belongs to sumx21t-3310/FloatSoda. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/sumx21t-3310/FloatSoda/main/.agents/skills/floatsoda-junior-coder-test/SKILL.md
Clone the repo
git clone --depth 1 https://github.com/sumx21t-3310/FloatSoda

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for floatsoda-junior-coder-test

README.md
[![agentmods](https://agentmods.dev/badge/skills/sumx21t-3310/floatsoda/floatsoda-junior-coder-test/github.svg)](https://agentmods.dev/skills/sumx21t-3310/floatsoda/floatsoda-junior-coder-test)
Your own site
<a href="https://agentmods.dev/skills/sumx21t-3310/floatsoda/floatsoda-junior-coder-test"><img src="https://agentmods.dev/badge/skills/sumx21t-3310/floatsoda/floatsoda-junior-coder-test/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for floatsoda-junior-coder-test

Your own site · 80×15
<a href="https://agentmods.dev/skills/sumx21t-3310/floatsoda/floatsoda-junior-coder-test"><img src="https://agentmods.dev/badge/skills/sumx21t-3310/floatsoda/floatsoda-junior-coder-test.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 309 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 3,773 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00309 $0.03773
Opus 5 $0.00154 $0.01886
Sonnet 5 $0.00062 $0.00755
Haiku 4.5 $0.00031 $0.00377

Measured yesterday against content hash d22be0734003, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade A, and why

floatsoda-junior-coder-test scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.agents/skills/floatsoda-junior-coder-test/SKILL.md · 160 lines

How it starts

The opening of the file, as written. The whole thing — 160 lines — stays where its author put it; the contents beside it link to each section on GitHub.

FloatSoda ジュニアコーダーテスト

このスキルが存在する理由

FloatSoda のペルソナ①は、ほとんどコードを書かない VRChatter — 代わりに LLM が書く。つまり docs と API サーフェスの本当の読者は LLM で、設計ゴールは「LLM がこの API を誤用できないこと」。 このスキルはそのゴールを運用に落とす: 低エフォートのモデルをそのジュニアコーダーの代役に 立て、公開 docs と現実的なタスクだけを渡し、つまずいた箇所を厳密にトリアージする。つまずきは すべてシグナル — docs が迷わせたか、ライブラリにバグがあるか、API が誤用しやすいかのいずれか。

これは実証済みのテスト: 初回の実行(Sonnet 5、medium)で実在のライブラリバグを2つ炙り出した — docs が推奨するフレームリミッターが全オーバーレイをクラッシュさせる件と、動的な子の削除での 無限再帰。実行ログは memory の vibe-coding-test-sonnet5-result にある。

テストを成立させる唯一のルール: ブラックボックス

ジュニアモデルは FloatSoda を NuGet の利用者とまったく同じように — docs だけを通して — 体験しなければならない。src/ は決して読ませない。実装を覗けてしまったら、テストはもはや docs/API の品質を測っておらず、それが目的のすべてだったはず。これはサブエージェントのプロンプト (後述)で強制し、作業ディレクトリを FloatSoda のリポジトリツリーの外に置くことでも強制する。

ゲートとしての2つの使われ方

このテストは場当たりな実行だけのものではない — 2つのプロセスゲートを支えている。ワークフローは 同じで、タスクの選び方(手順1)だけが異なる。

  • リリースゲート(RELEASING.md 参照): リリースのタグ付け前に、リリースコミット上で実行する。 mainhard のタスクを使い、リリースごとにテーマをローテーションする(トースト / フォトアルバム / FaceEmo)。サーフェスが1つのシナリオに過適合しないようにするため。docs バグ(ⓑ) かライブラリバグ(ⓒ)が出たらリリースをブロックする。
  • 新 API PR ゲート(CONTRIBUTING.md 参照): public で文書化される API を追加・変更する PR で実行する。新 API を使わないと完了できないタスクを書き、ジュニアには PR ブランチの 更新後 docs だけを渡す。Phase 2 の間は、該当 PR はすべて特定の Flutter ウィジェット 1つを移植するものなので、タスクは発明せずそのウィジェットの Flutter ドキュメントから導出 しなければならない — 必要な変形手順は references/prompts.md の「🧩 Phase 2」節を見る。 こちらの方がレバレッジが高い: 誤用されやすい API を、まだ変えられるうちに、public になって 固定される前に捕まえる。失敗は設計シグナルとして読む:
    • 見つけられない → 発見性 / docs のギャップ(WidgetSystem に無い、直感的でない名前)
    • 誤って使う → API が misuse-proof でないか、docs が曖昧
    • 別の形をハルシネーションする → モデルは実装より自然な API を書いた。実 API をその直感へ 寄せることを強く検討する。これはプロジェクトの「LLM がこの API を誤用できないこと」という ゴールの具体化であり、PR レビュー時が最も安く動ける時期。

ワークフロー

1. タスクを選ぶ

references/prompts.md から選ぶ(easy / main / hard、VRChatter の声で書かれている)か、同じ 精神で新しく書く。測りたいものに難易度を合わせる:

  • easy — getting-started の経路がそもそも機能するか(オーバーレイが1枚出るか)
  • main — 実装済みウィジェットだけを使う、状態を持つ動的な UI(スイートスポット)
  • hard — 既知のギャップ(hit-testing、未実装のスタブウィジェット)へ意図的に踏み込ませ、 モデルが API をハルシネーションするか、正しく諦めるか、回避策を見つけるかを観察する

Read the full file on GitHub · 160 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. yesterday Changed · +8 lines · +98 tokens per session d22be0734003
  2. 11d ago First seen · 152 lines · 211 tokens per session scan A 86614f279a3a

Subscribe to this mod's changes

floatsoda-junior-coder-test is a skill published in the GitHub repository sumx21t-3310/FloatSoda (11 stars, last pushed yesterday), licensed MIT. It adds 309 tokens to every session and 3,773 once invoked, about $0.0015 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories

author-test

Generate a test given sample. Parameters: C# SDK repository root; Package name: one of Azure.AI.Projects, Azure.AI.Projects.Agents or Azure.AI.Extensions.OpenAI; the sample to use as a starting point for the test.

Azure/azure-sdk-for-net · 68 tokens

starlark-dev

Develop and debug Kurtosis Starlark packages. Create packages from scratch, understand the plan-based execution model, use print() debugging, handle future references, and test packages locally. Use when writing or troubleshooting .star files.

kurtosis-tech/kurtosis · 50 tokens

migrate-mstest-v3-to-v4

Use this skill before answering, planning, or editing any MSTest 3.x-to-4.x upgrade or post-upgrade failure. Triggers include "MSTest v4 breaking changes"; CS0507/CS0103/CS1061/CS1615; ExecuteAsync, CallerInfo, DisplayName, or custom TestMethodAttribute; ClassCleanupBehavior; ContainsKey; ThrowsExactly or…

managedcode/dotnet-skills · 179 tokens

migrate-xunit-to-mstest

Convert .NET tests from xUnit.net v2/v3 to MSTest v4 while preserving VSTest or MTP. Use for replacing xunit packages, Fact/Theory/InlineData/MemberData, assertions, IClassFixture/ICollectionFixture, ITestOutputHelper, TestContext cancellation, traits/Owner, skips, timeouts, and xUnit parallelization. Also use when a…

managedcode/dotnet-skills · 141 tokens

migrate-xunit-to-xunit-v3

Migrate .NET test projects from xUnit.net v2 to xunit.v3 and fix v3 breaks. Use for package/CPM conversion, OutputType=Exe, preserving the VSTest or MTP runner (including projects currently using YTest.MTP.XUnit2), incompatible TFMs, async void tests, string-to-Type attributes, custom Fact/Theory/BeforeAfterTest…

managedcode/dotnet-skills · 149 tokens

semantic-kernel

Build AI-enabled .NET applications with Semantic Kernel using services, plugins, prompts, and function-calling patterns that remain testable and maintainable. USE FOR: adding AI-driven prompts, plugins, or orchestration to a .NET app; reviewing kernel construction, service registration, or plugin usage; building…

managedcode/dotnet-skills · 116 tokens