Guide · Agent skill
How to use the Team QA Testing Cycle agent skill (team-qa)
Updated · written with AI from the project's own documentation and checked against our catalogue data
team-qa is an agent skill for Claude that tells the agent to act as a QA team lead and tester across a full testing cycle: strategy, test plan, test cases, execution and sign-off. It is written for teams using the Claude Code Game Studios folder structure. After this guide you will know what an agent skill is, how to install this one, how to trigger it with a plain request, and what it needs before it runs.
Team QA Testing CycleDonchitos/Claude-Code-Game-Studios · MIT Download the skill Quick start
- 1Download the ZIP. Download the team-qa ZIP from this page.
- 2Unzip into a skills folder. Unzip it into ~/.claude/skills/ for personal use or into .claude/skills/ in your project.
- 3Or upload in the Claude apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the ZIP.
- 4Check the prerequisites. Confirm Claude Code has the Agent, AskUserQuestion and file tools enabled, and that the repo files the skill references are present.
- 5Make a natural request. Ask for a QA cycle and name the sprint or feature, for example sprint-03 or feature: combat.
- 6Answer the checkpoints. At each phase transition, pick the option that matches what you want the agent to do next.
- 7Review the sign-off report. Read the sign-off report and bug reports before treating the testing cycle as finished.
What is an agent skill, and what does team-qa make the agent do?
An agent skill is a folder containing a SKILL.md file with instructions that Claude loads when a task matches the skill. You do not run it like a normal program; you ask for the task and the agent follows the skill's process.
When team-qa is invoked, the skill tells the agent to plan a test strategy, write test cases and bug reports, walk through the testing phases and produce a sign-off report. At each phase transition it asks you to choose from options using AskUserQuestion, so you stay in control of when the work moves on.
- qa-lead role: strategy, test plan, story classification and sign-off report
- qa-tester role: test case writing, bug report writing and manual QA documentation
- Decision checkpoints between phases, with the user choosing the next step
- Depth of the process depends on settings read from a config file (team size, review mode, QA level)
Install the team-qa skill from the ZIP or the Claude apps
Option 1 (Claude Code, personal): download the team-qa ZIP from this page and unzip it into ~/.claude/skills/ so that the skill folder sits there with SKILL.md inside it.
Option 2 (Claude Code, single project): unzip the same ZIP into the .claude/skills/ folder of your project so the skill is available only in that project.
Option 3 (Claude apps): download the ZIP, then upload it in the Claude apps under Settings → Capabilities → Skills.
- Personal: ~/.claude/skills/
- Project only: .claude/skills/ inside your project folder
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
~/.claude/skills/team-qa/SKILL.mdTrigger team-qa with a natural request
The skill's argument hint lists two ways to scope a run: a sprint identifier such as sprint-03, or a feature named as feature: followed by the system name. You can add --review with full, lean or solo to set review depth.
You can also make the request in plain language that names the QA cycle, the sprint or the feature. Mention the scope so the agent knows which story files to load.
- Plan QA for sprint-03 using the team-qa skill
- Run a team-qa testing cycle for feature: combat, with --review lean
- Write test cases and a sign-off report for the inventory feature
[sprint | feature: system-name] [--review full|lean|solo]What does team-qa need before it runs?
The skill is a process guide. It depends on other files and tools in the wider Claude Code Game Studios repo. Those files are not included in the SKILL.md you download, so the skill will not work fully on its own.
The skill's allowed tools are Read, Glob, Grep, Write, Agent, AskUserQuestion and one restricted Bash command. Its Phase 0 reads config through a helper script and expects the referenced documents to be present.
- Claude Code with the Agent, AskUserQuestion and file tools enabled
- hooks/yaml-helper.sh from the wider repo
- Referenced docs: .claude/docs/automation-modes.md, config-resolution.md and coding-standards.md
- Bash access limited to running yaml-helper.sh resolve_config
- Story and sprint files in production/ for the scope you name
Safety notes: scripts and network access
This skill runs a script. Its only permitted Bash call is the yaml-helper.sh resolve_config command, which reads configuration. Review that script in the repo before you rely on the skill in a shared environment.
The skill does not use the network, according to its listing. It writes its outputs as files, such as test plans, test cases and sign-off reports, under production/, docs/ or tests/ paths it names in the run.
- Runs scripts: yes (one restricted yaml-helper.sh call)
- Uses network: no
- Writes new artifacts under production/, docs/ or tests/ rather than editing existing source or config
- Check the outputs before you accept a sign-off report
Tips for getting useful output from team-qa
Pick the smallest scope that answers your question. A single feature or sprint keeps the test plan readable. Use the review option to match the depth you want, and answer each checkpoint with a clear choice, because the next phase depends on it.
Expect the process to add overhead. For a quick bug check or a small script, a direct request to the agent is usually faster.
- Start with one feature before running a whole sprint
- Use the decision checkpoints to stop a phase you do not agree with
- Keep story files tagged by system so feature scope can be found
- Read the sign-off report as a draft until a person has reviewed it
What people use it for
- Planning a test strategy and test plan for a sprint in a game studio project
- Writing manual test cases and bug reports for a single feature
- Working through testing phases with a lead's decision checkpoints before each move
- Producing a sign-off report at the end of a sprint or feature cycle
Frequently asked questions
Is team-qa free to use?
The skill is listed under the MIT licence on its source repository. The licence covers the code in the repo, so you can check the licence text there for the exact terms.
How do I install the team-qa skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use or .claude/skills/ inside a project. In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.
What do I need to run team-qa?
You need Claude Code with the Agent, AskUserQuestion and file tools enabled. You also need the referenced files from the wider repo, including hooks/yaml-helper.sh and the docs in .claude/docs/, because they are not included in the SKILL.md file alone.
Does team-qa run tests automatically?
No. The skill describes the workflow and delegates work to agents. It does not execute your software tests by itself, and any real test runs depend on other files and tools.
Can I use team-qa outside a Claude Code Game Studios project?
The skill depends on referenced docs, hooks and agents from the Claude Code Game Studios folder structure. Projects outside that structure are not a good fit unless you supply those files.
Why is team-qa slower than asking the agent for a quick test?
The skill runs a multi-phase process with checkpoints and written outputs. That adds overhead, so for a one-off bug check or a small script a direct request is usually more practical.
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
Use team-qa if you work in a Claude Code Game Studios style project and want a structured testing cycle with decision checkpoints, test cases and a sign-off report for a sprint or feature. Pick something else for quick bug checks, small scripts, or projects outside that folder structure, and do not expect it to run tests automatically from this file alone.
Download the skill