Agent skill

Tiny Game Controller

by minigdx in minigdx/tiny

Controls a running Tiny game engine over its local debug HTTP server, sending key presses, releases and taps so the agent can play or test the game.

MITAuto-check passedGame Development

Install Tiny Game Controller

skills CLI
$ npx skills add minigdx/tiny --skill game-control -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install minigdx/tiny game-control --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/minigdx/tiny.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/game-control .claude/skills/game-control && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
game-control
GitHub stars
164
Token cost
~874 tokens
SKILL.md length
284 words
Files
1
Skills in repo
1
Repo updated
First seen
Licence
MIT

At a glance

Controls a running Tiny game engine over its local debug HTTP server, sending key presses, releases and taps so the agent can play or test the game.

  • Works in 4 steps: If arguments describe an action (e.g.,… → If arguments are a key name (e.g.,… → If no arguments or "help" → …
  • Playing or exploring a running Tiny game through the agent
  • SKILL.md covers Server, Available Endpoints, Available Keys and Usage Patterns, plus 3 more sections
  • Calls curl

What it does

The skill talks to the debug server that Tiny starts with its CLI run command, by default on localhost port 8081. Four endpoints cover the work: listing available keys, pressing a key that stays down until released, releasing it, and tapping, which presses and releases after a short delay. Keys include the arrows, space, enter, escape, letters, number keys, modifiers and function keys.

The agent translates a plain instruction such as move right, jump or press space three times into curl calls, with sleeps for held keys, repeated taps for loops and two presses at once for diagonal movement. A bare key name means a single tap, and no argument lists the keys with usage examples. For open-ended requests like play the game, it takes a screenshot first, through the serve endpoint or Playwright, sends inputs based on what it sees and repeats. Allowed tools are limited to curl, sleep, file reading and three Playwright browser calls.

When your agent uses it

  • Playing or exploring a running Tiny game through the agent
  • Testing game input handling with scripted key presses
  • Tapping or holding keys to reproduce a bug in a game level

Example prompts

  • “Move the character right for half a second, then jump.”
  • “Press space three times and take a screenshot of the result.”
  • “List the keys the running game accepts.”

Requirements

  • A Tiny game running via tiny-cli with its debug server on port 8081
  • curl
  • Pre-approved tools (allowed-tools): Bash(curl:*), Bash(sleep:*), Read, mcp__playwright__browser_take_screenshot, mcp__playwright__browser_navigate, mcp__playwright__browser_snapshot

Workflow steps

4 steps, taken from the first numbered list in SKILL.md.

  1. If arguments describe an action (e.g., "move right", "jump", "press space 3 times")
  2. If arguments are a key name (e.g., "SPACE", "ARROW_LEFT")
  3. If no arguments or "help"
  4. If "play" or complex instructions (e.g., "play the game", "explore the level")

What it can do on your machine

Read from SKILL.md and the folder at commit 876365c. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves these tools, so the agent can use them without asking each time:

    • Bash(curl:*)
    • Bash(sleep:*)
    • Read
    • mcp__playwright__browser_take_screenshot
    • mcp__playwright__browser_navigate
    • mcp__playwright__browser_snapshot

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • curl

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use curl, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Tiny Game Controller loads about 874 tokens when it runs. Until then it costs about 50 tokens; SKILL.md has 284 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~50
When it runs · the whole SKILL.md, loaded when a task matches
~874

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from minigdx/tiny at commit 876365c, republished under its MIT licence (© minigdx). 284 words, ~874 tokens.

Download SKILL.mdSave it as .claude/skills/game-control/SKILL.md (or your agent's skills folder).
name
game-control
description
Control a running Tiny game engine via HTTP. Send key presses, releases, and taps to interact with the game. Use when the user asks to play, control, test, or interact with a running game.
allowed-tools
Bash(curl:*), Bash(sleep:*), Read, mcp__playwright__browser_take_screenshot, mcp__playwright__browser_navigate, mcp__playwright__browser_snapshot
argument-hint
[action] [key or instructions]

Game Control Skill

You can control a running Tiny game engine by sending HTTP requests to the debug server.

Server

The game debug server runs on http://localhost:8081 by default (started with tiny-cli run).

Available Endpoints

List available keys
bash
curl -s http://localhost:8081/control/keys
Press a key (stays pressed until released)
bash
curl -s -X POST "http://localhost:8081/control/press?key=KEY_NAME"
Release a key
bash
curl -s -X POST "http://localhost:8081/control/release?key=KEY_NAME"
Tap a key (press + auto-release after ~50ms)
bash
curl -s -X POST "http://localhost:8081/control/tap?key=KEY_NAME"

Available Keys

Direction: ARROW_LEFT, ARROW_RIGHT, ARROW_UP, ARROW_DOWN Action: SPACE, ENTER, ESCAPE, TAB, BACKSPACE, DELETE Letters: A through Z Numbers: NUM0 through NUM9 Modifiers: SHIFT, CTRL, ALT Function: F1 through F12

Usage Patterns

Tap a key once (one-shot action like jump or shoot)
bash
curl -s -X POST "http://localhost:8081/control/tap?key=SPACE"
Hold a key for sustained movement
bash
# Hold right for 500ms then release
curl -s -X POST "http://localhost:8081/control/press?key=ARROW_RIGHT"
sleep 0.5
curl -s -X POST "http://localhost:8081/control/release?key=ARROW_RIGHT"
Multiple rapid taps
bash
for i in $(seq 1 5); do
  curl -s -X POST "http://localhost:8081/control/tap?key=SPACE"
  sleep 0.1
done
Simultaneous keys (e.g., diagonal movement)
bash
curl -s -X POST "http://localhost:8081/control/press?key=ARROW_RIGHT"
curl -s -X POST "http://localhost:8081/control/press?key=ARROW_UP"
sleep 0.5
curl -s -X POST "http://localhost:8081/control/release?key=ARROW_RIGHT"
curl -s -X POST "http://localhost:8081/control/release?key=ARROW_UP"

How to Use

When the user provides $ARGUMENTS:

  1. If arguments describe an action (e.g., "move right", "jump", "press space 3 times"):

    • Translate the instruction into the appropriate curl commands
    • Execute them
  2. If arguments are a key name (e.g., "SPACE", "ARROW_LEFT"):

    • Tap that key once
  3. If no arguments or "help":

    • List the available keys by calling GET /control/keys
    • Show usage examples
  4. If "play" or complex instructions (e.g., "play the game", "explore the level"):

    • Take a screenshot first to see the game state (if possible via the serve endpoint or playwright)
    • Send appropriate inputs based on what you observe
    • Iterate: act, observe, act again

Error Handling

  • If curl fails with connection refused, the game is not running. Tell the user to start it with tiny-cli run.
  • If the response contains "error":"Engine not ready", the game is still loading. Wait a moment and retry.
  • If the response contains "error":"Unknown key", check the key name against the available keys list.

Response Format

All endpoints return JSON:

  • Success: {"ok":true,"action":"press","key":"ARROW_LEFT"}
  • Error: {"error":"Missing 'key' query parameter"}

© minigdx, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .claude/skills/game-control of minigdx/tiny.

Open the folder on GitHubat commit 876365c

Compare with similar skills

Tiny Game Controller next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Tiny Game Controller compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Tiny Game Controller this skillminigdx/tiny164—~874Automated safety check: PassMIT
Game Boy ROM Playthrough Drivertrekawek/coffee-gb1.2k—~1.3kAutomated safety check: PassMIT
Ship Web Gamesnirholas/three.ws2271 repos~284Automated safety check: PassApache-2.0
Evidence-Based Game QAzenstory-ai/novel-to-game834—~389Automated safety check: PassMIT
Build Game Inventorynirholas/three.ws2271 repos~418Automated safety check: PassApache-2.0
Test Playable Web Gamesnirholas/three.ws2271 repos~429Automated safety check: PassApache-2.0

Similar skills

  • Plays an authorized Game Boy or Game Boy Color ROM in one persistent headless Coffee GB session, inspecting frames and keeping an action trace for replay or tests.

    1.2k GitHub stars~1.3k tokensUpdated 7 days ago
    Game DevelopmentAuto-check passed
  • Ship Web Games

    nirholas/three.ws

    Package, deploy, and verify a playable Three.js or web game.

    227 GitHub starsUsed in 1 repo~284 tokens
    Game DevelopmentAuto-check passed
  • Evidence-Based Game QA

    zenstory-ai/novel-to-game

    Verifies a game build on its target runtime with real run evidence for launch, rendering, input, core loop, outcome and restart, and records limitations honestly.

    834 GitHub stars~389 tokensUpdated 5 days ago
    Game DevelopmentAuto-check passed
  • Build Game Inventory

    nirholas/three.ws

    Build or repair game inventory, loot, equipment, tooltips, drag-and-drop, persistence, and progression systems.

    227 GitHub starsUsed in 1 repo~418 tokens
    Game DevelopmentAuto-check passed
  • Test Playable Web Games

    nirholas/three.ws

    Test a playable browser game end to end with deterministic fixtures and real browser evidence.

    227 GitHub starsUsed in 1 repo~429 tokens
    Testing & QAAuto-check passed
  • Investigates a stubbornly failing Playwright test as a possible product bug, using error output, screenshots, traces and server code, and writes a structured bug report.

    41k GitHub stars~1.5k tokensUpdated today
    Testing & QAAuto-check passed

Works with

Questions about Tiny Game Controller

What does Tiny Game Controller do?

Controls a running Tiny game engine over its local debug HTTP server, sending key presses, releases and taps so the agent can play or test the game. The skill talks to the debug server that Tiny starts with its CLI run command, by default on localhost port 8081. Four endpoints cover the work: listing available keys, pressing a key that stays down until released, releasing it, and tapping, which presses and releases after a short delay.

When should I use Tiny Game Controller?

Tiny Game Controller fits situations like: playing or exploring a running Tiny game through the agent; testing game input handling with scripted key presses; tapping or holding keys to reproduce a bug in a game level.

How do I install Tiny Game Controller in Claude Code?

Run `npx skills add minigdx/tiny --skill game-control -a claude-code`. Or copy the skill folder (.claude/skills/game-control in minigdx/tiny) into .claude/skills/game-control in your project. Claude Code loads it when a task matches its description.

How do I install Tiny Game Controller in Codex?

Run `npx skills add minigdx/tiny --skill game-control -a codex`. Or copy the skill folder (.claude/skills/game-control in minigdx/tiny) into .agents/skills/game-control in your project. Codex loads it when a task matches its description.

Can I use Tiny Game Controller in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add minigdx/tiny --skill game-control -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/game-control, .gemini/skills/game-control, .github/skills/game-control and .opencode/skills/game-control in your project.

What does Tiny Game Controller need to run?

Going by SKILL.md and its folder, Tiny Game Controller needs the command-line tools its instructions call (curl). Our summary lists: A Tiny game running via tiny-cli with its debug server on port 8081; curl. Its frontmatter pre-approves these tools: Bash(curl:*), Bash(sleep:*), Read, mcp__playwright__browser_take_screenshot, mcp__playwright__browser_navigate, mcp__playwright__browser_snapshot.

Does Tiny Game Controller access the network?

SKILL.md contains no URLs. Its commands use curl, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Tiny Game Controller safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Tiny Game Controller use?

Tiny Game Controller is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Tiny Game Controller use?

About 874 tokens (SKILL.md is roughly 3.5k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Tiny Game Controller?

Skills that share tags, products or a category with Tiny Game Controller: Game Boy ROM Playthrough Driver (trekawek/coffee-gb, 1.2k stars), Ship Web Games (nirholas/three.ws, 227 stars), Evidence-Based Game QA (zenstory-ai/novel-to-game, 834 stars) and Build Game Inventory (nirholas/three.ws, 227 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Tiny Game Controller?

minigdx (a GitHub organization) maintains it in minigdx/tiny, which has 164 GitHub stars. The repository was last updated on May 28, 2026.

Source: minigdx/tiny on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.