Bounded Engine Runs logo

Bounded Engine Runs

Organization
SummerEngine
bounded-engine-runs

Run Summer headless or offscreen safely: time limits, one engine per project, import passes, crash checks.

Overview

PublisherSummerEngine
Repositorysummer
Skill namebounded-engine-runs
Stars
91
Forks
8
Bundled files
1
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.

  • 1 bundled files

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

  • Open source

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

Installation

Install the Bounded Engine Runs 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/SummerEngine/summer/tree/main/library/skills/bounded-engine-runs
  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/SummerEngine/summer.git /tmp/summer
mkdir -p .claude/skills
cp -r /tmp/summer/library/skills/bounded-engine-runs .claude/skills/bounded-engine-runs
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Bounded Engine Runs 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 Bounded Engine Runs 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 Bounded Engine Runs 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.

Bounded engine runs for agents

Agents verify games by running the engine many times: imports, tests, captures, multiplayer bots. A run that hangs, overlaps another, makes noise or hides an error costs time. These rules keep runs safe when several agents share one project on one machine.

1. Every run is bounded, muted and alone

  • Bound it. macOS has no timeout. Wrap runs so the engine itself gets killed:
    sh
    perl -e 'alarm shift; exec @ARGV' 180 <summer> --headless --path . -s res://tests/run_tests.gd
    alarm survives exec; exit code 142 means the limit hit.
  • One engine per project at a time. Two engines writing the same .godot/ import cache corrupt it or freeze each other's captures. Serialize with a lock file held by the engine process:
    sh
    perl -e 'use Fcntl qw(:flock); $^F=255; open(my $l,">>",shift) or die; flock($l,LOCK_EX); alarm shift; exec @ARGV' \
      <project>/.engine.lock 180 <summer> ...
    Put this in one wrapper script and make every agent use it.
  • On Windows there is no perl by default, and Summer.exe writes its output only to --log-file. Start it with Start-Process -PassThru, wait with WaitForExit, kill it on the limit, and hold the lock file open for the whole run:
    powershell
    $lock = [IO.File]::Open("<project>\.engine.lock", 'OpenOrCreate', 'ReadWrite', 'None')   # throws while another run holds it; retry later
    $p = Start-Process <summer.exe> -ArgumentList '--headless --path . -s res://tests/run_tests.gd --log-file run.log' -PassThru -NoNewWindow
    if (-not $p.WaitForExit(180000)) { $p | Stop-Process; 'timed out' }
    $lock.Close()
  • Cap its memory. A time limit does not stop a run from eating all memory first; several engine runs plus a browser loading a large web export can exhaust memory and crash the machine. The wrapper should run the command as a child, sum the resident memory of its whole process tree (browsers spawn GPU/renderer helpers), kill the tree past a cap (e.g. 3 GB), and also stop when system free memory falls low (sysctl -n kern.memorystatus_level < 15 on macOS), because GPU memory on Apple Silicon is system memory but is not counted in RSS. Refuse to start when the system is already short (< 35).
  • Mute it. --audio-driver Dummy.
  • Render only when you must, and offscreen. --headless for logic (no pixels). For visual checks use --summer-offscreen with a fixed --resolution and a capture script that saves PNGs and quits. Visible windows only when a human asked to watch.

2. Imports and parse errors

  • New class_name scripts are unknown to -s scripts until an import pass registers them: <summer> --headless --audio-driver Dummy --path . --import.
  • The first import may rewrite project.godot (reorder keys, add a [summer] section; one trial saw a hand-written [rendering] section dropped). Check git diff project.godot after it, and pass renderer choices on the command line (--rendering-method ...) when a run depends on them.
  • One broken class_name script cascades into "Could not parse global class" and dozens of unrelated "cannot infer type" errors. Get the real line with: <summer> --headless --path . --check-only -s res://path/to/file.gd
  • Compile every script before booting the game: a tiny SceneTree script that load()s each .gd under a folder and reports can_instantiate() failures, exit code = failures.

3. Tests that can't lie

  • A runtime script error inside a test does not fail it by default; the test just stops and counts as passed. Register a logger and fail the test when its error count rises:
    gdscript
    class ErrorCounter extends Logger:
    	var script_errors := 0
    	func _log_error(_f: String, file: String, _line: int, _c: String, _r: String, _n: bool, type: int, _bt: Array[ScriptBacktrace]) -> void:
    		if type == Logger.ERROR_TYPE_SCRIPT or file.ends_with(".gd"):
    			script_errors += 1
    	func _log_message(_m: String, _e: bool) -> void:
    		pass
    # in the runner: OS.add_logger(counter); compare counter.script_errors before and after each test
  • Exit 0 is not success. Always grep the log for SCRIPT ERROR, Parse Error and ERROR:, and print one explicit success line (TESTS passed=N failed=0) you can check for.

4. Typed GDScript traps agents hit repeatedly

  • var x := dict["k"] cannot infer a type (Dictionary values are Variant). Write var x: int = dict["k"].
  • A loop variable that shadows an outer variable is a parse error that takes the whole class down.
  • Don't name a class after an engine class (SummerSession); the error only says it "hides a native class".

5. Web exports on a dev machine

  • Check the .pck size before opening it in a browser; the browser holds all of it in memory. A project that exports every imported asset of a large scene can pass 250 MB. Ship only what the game uses first.
  • Load it in a browser only under the same memory-capped wrapper, one at a time, never next to other engine runs.

6. Captures you can trust

  • Read every PNG you produce. If several files are identical (md5 -r *.png), the run froze; rerun alone.
  • Keep evidence outside the project (logs, PNGs) so imports never pick it up. If it must live inside, put an empty .gdignore file in that folder so the editor and imports skip it.

See also

  • headless-scripting (writing the SceneTree scripts these runs execute)
  • multiplayer-testing (Local Play bot runs)
  • compatibility-renderer-traps (dual-renderer captures)

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 Bounded Engine Runs AI skill do?

Run Summer headless or offscreen safely: time limits, one engine per project, import passes, crash checks.

Why use Bounded Engine Runs on TypingMind?

Because you install it once and use it with any model. Bounded Engine Runs 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 Bounded Engine Runs in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/SummerEngine/summer/tree/main/library/skills/bounded-engine-runs. 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 Bounded Engine Runs?

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 Bounded Engine Runs?

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

Is the Bounded Engine Runs AI skill free?

Yes. It is published on GitHub by SummerEngine 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 👇