Guide · Agent skill
How to use the Team Combat Orchestrator agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
Team Combat is an agent skill for Claude Code that asks the agent to work through a combat feature the way a game studio combat team would: design first, then implementation, then validation. It is built for game developers using the Claude Code Game Studios framework, and for solo developers who want a structured design-to-test checklist for a single mechanic. By the end of this guide you will know what the skill does, how to install it, how to trigger it with a request, and what it depends on before you try it.
Team Combat OrchestratorDonchitos/Claude-Code-Game-Studios · MIT Download the skill Quick start
- 1Confirm the prerequisites. Check that you have Claude Code or a compatible runtime, a game engine project, and the Claude Code Game Studios repository files, including hooks/yaml-helper.sh and .claude/docs.
- 2Download and unzip the skill. Download the team-combat ZIP from this page and unzip it into ~/.claude/skills/ for personal use, or into the project's .claude/skills/ folder for one project.
- 3Or upload it in the Claude apps. In the Claude apps, open Settings, go to Capabilities, then Skills, and upload the ZIP.
- 4Invoke the skill with a feature description. Run the slash command with a concrete combat mechanic, such as a melee parry system or ranged weapon spread, and an optional --review setting.
- 5Make your phase decisions. Read the agent's analysis at each transition, then choose one of the options it presents to move the pipeline forward.
- 6Check the validation output. Review the QA tester's test cases and validation results before treating the feature as done.
What is an agent skill, and what does team-combat do?
An agent skill is a folder containing a SKILL.md file that tells an AI agent how to handle a specific kind of task. When you invoke a skill, the agent follows the instructions in that file. The skill is typically triggered by a request or a slash command.
When invoked with a combat feature description, team-combat tells the agent to run a structured pipeline that mirrors a game studio combat team. The pipeline moves through design, implementation and validation phases. At each phase transition the agent asks you to choose between options. It also hands parts of the work to role-specific subagents: a game designer, gameplay programmer, AI programmer, technical artist, sound designer and QA tester.
- Design: the game designer defines the mechanic, formulas and edge cases.
- Implementation: the gameplay programmer builds the core code, with the AI programmer for enemy behaviour where the feature needs it.
- Visual and audio: the technical artist covers VFX and visual feedback, and the sound designer covers audio events.
- Validation: the QA tester writes test cases and checks the implementation.
Install the team-combat skill
Download the team-combat ZIP from this page. You can install it in one of two ways, depending on where you use the agent.
For Claude Code, unzip the download into your personal skills folder at ~/.claude/skills/, or into a project's .claude/skills/ folder if you want the skill limited to that project. For the Claude apps, open Settings, go to Capabilities, then Skills, and upload the ZIP there.
- Personal use across projects: ~/.claude/skills/
- Single project only: the .claude/skills/ folder inside that project
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
How to trigger the skill with a natural request
The skill is user-invocable and takes a combat feature description as its argument. The documented usage is a slash command with an optional review setting. Without a description, the skill prints a usage message and stops without spawning any subagents or reading files.
Describe a concrete mechanic and its context. Short, specific descriptions produce a more focused design than vague ones. The skill's own examples are a melee parry system and a ranged weapon spread.
- Give the feature and the game context, for example: design a melee parry system for our action RPG.
- Name the review depth if you want one: full, lean or solo.
- Expect the agent to pause and ask you to choose options at each phase transition.
/team-combat [combat feature description] [--review full|lean|solo]Understand the phases and your decision points
The pipeline pauses at phase transitions and presents the subagent proposals as selectable options. The agent writes its full analysis in the conversation first, then records your decision with short labels. How much you must approve depends on the automation setting in the project's config: in collaborative mode you approve each phase before the next begins, while in guided or autonomous mode the pipeline advances on its own unless a phase is blocked.
The team.size setting decides which agents actually run. At the default individual size, the gameplay programmer runs the pipeline and the AI programmer joins only if the feature involves AI. Larger sizes activate more of the team. The skill announces which agents it will spawn before Phase 1 begins, so you can see the shape of the run.
- Collaborative: you approve every phase before the pipeline continues.
- Guided: the pipeline advances unless a phase is BLOCKED.
- Autonomous: the pipeline runs end to end and logs each phase outcome.
Safety, requirements and what to check first
The skill runs scripts and uses no network access. Its allowed tools include a Bash command restricted to the resolve_config step of the framework's hooks/yaml-helper.sh script. It reads and writes project files through the standard file tools and uses the Agent tool to spawn subagents. Review the SKILL.md file before first use, since it runs a helper script from the wider framework.
The skill needs more than its own folder. It requires Claude Code or a compatible runtime that supports SKILL.md skills and subagents, the Claude Code Game Studios repository files (including hooks/yaml-helper.sh and the .claude/docs configuration), and a game engine project. Where configured, it uses an engine specialist agent for Godot, Unity or Unreal.
- Runs scripts: yes, via a restricted Bash command for the config helper.
- Uses network: no.
- Without the framework's hook, config and docs files, the referenced helper and config will be missing and the pipeline will not run as written.
What people use it for
- Designing a melee parry system with defined formulas, edge cases and acceptance criteria before any code is written.
- Planning a ranged weapon spread mechanic, including how its enemy AI and visual feedback should behave.
- Producing a design-to-test checklist for one combat mechanic as a solo developer working in an existing engine project.
- Running a staged combat feature build inside a Claude Code Game Studios project where the framework's roles and config are already set up.
Frequently asked questions
How do I install the team-combat agent skill?
Download the team-combat ZIP from this page and unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder for a single project. In the Claude apps, you can instead upload the ZIP under Settings, Capabilities, then Skills.
What do I need to run team-combat?
You need Claude Code or a compatible runtime that supports SKILL.md skills and subagents, a game engine project, and the Claude Code Game Studios repository files. The skill relies on hooks/yaml-helper.sh and the .claude/docs configuration, which are not included in the skill folder.
Is team-combat free to use?
The skill is published under the MIT licence. Check the repository's licence file for the full terms before you use or redistribute it.
Can I use team-combat for a non-game project?
No. The pipeline and its roles, including the game designer, gameplay programmer and sound designer, are specific to game development, so it is not intended for non-game projects.
Why does the skill fail or not run as written?
The most likely cause is missing framework files. The skill calls hooks/yaml-helper.sh and reads config from .claude/docs, and if those are not present the referenced helper and config will be missing and the pipeline will not run as written. Install the full Claude Code Game Studios repository files first.
How does team-combat compare with using a single prompt?
Team-combat adds a fixed structure: it moves through design, implementation and validation, asks you to choose at each transition, and delegates to role-specific subagents. If you only need a quick one-off answer for a mechanic, a plain prompt is simpler. The skill suits work where you want staged decisions and a repeatable checklist.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Team Combat suits game developers already using the Claude Code Game Studios framework, and solo developers who want a staged design-to-test checklist for one mechanic. It is a poor fit for non-game projects or for anyone without the framework's hook, config and docs files, since the pipeline will not run as written without them. Read SKILL.md and check its script access before you run it.
Download the skill