Uloop Execute Dynamic Code logo

Uloop Execute Dynamic Code

Community
hatayama
uloop-execute-dynamic-code

Execute C# with Unity APIs when existing uloop tools cannot inspect or edit enough. Use for reachable scene/component state, scene/prefab/menu automation, and PlayMode checks

Overview

Publisherhatayama
Repositoryunity-cli-loop
Skill nameuloop-execute-dynamic-code
Stars
578
Forks
49
Bundled files
2
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 2 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by hatayama on GitHub. Read the source before you install it.

Installation

Install the Uloop Execute Dynamic Code AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/hatayama/unity-cli-loop/tree/main/.agents/skills/uloop-execute-dynamic-code
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/hatayama/unity-cli-loop.git /tmp/unity-cli-loop
mkdir -p .claude/skills
cp -r /tmp/unity-cli-loop/.agents/skills/uloop-execute-dynamic-code .claude/skills/uloop-execute-dynamic-code
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Uloop Execute Dynamic Code in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Uloop Execute Dynamic Code on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Uloop Execute Dynamic Code is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Task

Run focused C# snippets in the active Unity Editor with uloop execute-dynamic-code.

For basic selected GameObject discovery or property inspection, use find-game-objects --search-mode selected first. Use this tool when the built-in inspection tools are not enough or when you need to modify Unity state.

This tool can inspect reachable Unity state — GameObjects, components, public properties, static values, method results — but it cannot read local variables or intermediate calculations inside an already-running method. When those values matter, follow the uloop-pause-point skill: a pause point's CapturedVariables carries the locals, parameters, and instance fields at that line with no code edit or recompile, and while Unity stays paused UloopPausePoint.TryGetCapturedValue(name) gives this tool live captured references (it returns (bool Found, object Value); see its references/captured-variables.md). That skill also covers the reverse combination — registering an EditorApplication.update watcher from this tool that freezes Unity on the first frame a runtime condition holds. Never poll or sleep inside a snippet; the body runs synchronously on the main thread.

Live state injection: when a running PlayMode session is merely in the wrong state — a stuck end-to-end scenario, a camera angle a raycast can never hit, a private flag blocking the path under test — fix the state instead of the code: write the field directly from this tool (reflection reaches private fields) and steer the session onward. Nothing recompiles and no domain reload happens, so the session's in-memory state survives intact. The snippet is a one-off diagnostic that never lands in source files, so project rules restricting reflection in production code are not violated.

Parameters

ParameterTypeDefaultDescription
--codestring-Inline C# statements to execute. Direct statements only; return is optional, and using directives may appear at the top of the snippet.
--parametersobject-Shell-quoted JSON object literal for reusing a snippet with varying data or keeping values outside the code. Values are exposed as parameters["param0"], parameters["param1"], and so on. Omit for most snippets; never pass a JSON string value.
--wait-for-domain-reloadflag-Wait for Domain Reload recovery after snippets that intentionally trigger Unity script reload or import work. Omit for normal inspection and editor-state workflows.
--yield-to-foreground-requestsflag-Allow foreground requests to preempt this execution

CLI-only flag, accepted instead of a schema parameter:

  • --code-file <path>: Read the C# statements from a file instead of --code. Use this when the active shell or launcher cannot preserve inline code exactly. Exactly one of --code or --code-file is required; combining them is an error.

Code Rules

Write direct statements from your own Unity API knowledge — no class/namespace/method wrappers. Return is optional. await works, and a returned Task is awaited: its value becomes Result, its exception a failure.

csharp
using UnityEngine;
float x = Mathf.PI;
return x;

Prefer terminal commands for file operations.

A type an active hot reload introduced is nameable here; write its full name. Members hot reload added to a type are not; see the hot-reload skill's added-field-wiring.md.

This snippet runs in the Editor execution context: Screen.width and Screen.height are Editor pixels, not the Game View resolution. For a ray through the Game View center, use cam.ViewportPointToRay(new Vector3(0.5f, 0.5f, 0f)).

Known transpiler constraints

On CS8421, CS8820, or a surprising type error around numeric literals (e.g. Color32 needing (byte)255), read references/transpiler-constraints.md — snippets hoist literals in ways that constrain static local functions and lambdas.

Shell Quoting

In zsh/bash, single-quote the whole snippet so C# double quotes pass through unchanged: --code 'return "hi";'. If a snippet fails to parse, gets mangled by the shell, or you are on Windows/PowerShell, read references/shell-quoting.md — or switch to --code-file.

When To Use Input Simulation Tools Instead

Calling UI handlers or runtime methods directly from a snippet is the better choice for targeted automation, direct state control, or quick diagnostics. Switch to the dedicated input tools only when the input route itself is part of what you need to verify:

ScenarioRecommended toolWhy
Verify that a uGUI element responds through the real EventSystem pointer pathsimulate-mouse-uiFires PointerDown / PointerUp / PointerClick / drag events through EventSystem raycasts instead of bypassing the UI input route.
Test gameplay that reads Mouse.current, button state, delta, or scrollsimulate-mouse-inputInjects Input System mouse state into Mouse.current so game code observes it like player input. Requires the New Input System (Input System Package (New) or Both); when that is unavailable, prefer an execute-dynamic-code workaround instead of changing project settings just to use the tool.
Jump straight to a known callback, invoke a method, inspect state, or set up a test preconditionexecute-dynamic-codeDirect automation without reproducing the full input pipeline.
Drive custom runtime behavior that does not map cleanly to the built-in input toolsexecute-dynamic-codeCalls project-specific methods and prototypes one-off flows immediately.

Output

Returns JSON:

  • Success: boolean — overall execution success
  • Result: string — value of the snippet's return statement (empty when omitted)
  • Logs: string[] — execution messages from the dynamic-code tool; read Unity Console Debug.Log output with get-logs
  • PartialResults (object, optional): values that a snippet explicitly saves before it completes or fails
  • CompilationErrors: object[] — Roslyn diagnostics with Message, Line, Column, ErrorCode, optional Hint and Suggestions
  • Error / ErrorMessage: string — top-level failure summary (empty on success)
  • UpdatedCode: string|null — the wrapped form actually compiled (handy when debugging using-statement reordering)
  • DiagnosticsSummary: string|null — compact summary when diagnostics are available
  • Diagnostics: object[] — structured diagnostics; same shape as CompilationErrors, usually populated together with it
  • Warning (string, optional): Set when Play Mode is running while the Unity Editor is unfocused. Progress may be throttled; run uloop focus-window, or use the pause-point --await/--trigger flow instead of polling for progress.

To retain an intermediate value across a later exception, opt in before the risky code:

csharp
UloopDynamicCodePartialResults.Set("completedSteps", completedSteps);

PartialResults contains only values explicitly saved this way. Ordinary local variables cannot be recovered after an exception unwinds the snippet. A cancellation before the snippet produces an execution result also returns no PartialResults.

On Success: false, inspect CompilationErrors first. If empty, read ErrorMessage (and Logs for extra context) — the failure may be a runtime exception, cancellation, or an "execution in progress" rejection, all of which return empty CompilationErrors. Both EditMode and PlayMode are supported — the snippet runs in whichever mode the Editor is in.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Uloop Execute Dynamic Code AI skill do?

Execute C# with Unity APIs when existing uloop tools cannot inspect or edit enough. Use for reachable scene/component state, scene/prefab/menu automation, and PlayMode checks

Why use Uloop Execute Dynamic Code on TypingMind?

Because you install it once and use it with any model. Uloop Execute Dynamic Code is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Uloop Execute Dynamic Code in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/hatayama/unity-cli-loop/tree/main/.agents/skills/uloop-execute-dynamic-code. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Uloop Execute Dynamic Code?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Uloop Execute Dynamic Code?

As many as you like. As long as a model supports skills, you can use Uloop Execute Dynamic Code with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Uloop Execute Dynamic Code AI skill free?

Yes. It is published on GitHub by hatayama under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇