Guide · Agent skill
How to use the Game Research Web Search agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The game-research-websearch skill is a set of instructions that teaches an AI agent to research a game, its engine or a modding technique in a systematic way. It is built for modders working on older games whose forums have gone offline, and for contributors who need every claim tied to a source URL. After reading this guide you will know how to install the skill, how to ask for it with natural requests, and what it will and will not do.
Game Research Web Searchrehan-remade/universal-modder · MIT Download the skill Quick start
- 1Download the ZIP. Download the ZIP for game-research-websearch from this page.
- 2Unzip into a skills folder. Unzip it so the game-research-websearch folder with SKILL.md sits in ~/.claude/skills/ or a project's .claude/skills/ folder.
- 3Or upload in the Claude apps. In the Claude apps, open Settings → Capabilities → Skills and upload the ZIP.
- 4Confirm internet and tools. Make sure you have internet access, the um command-line tool and yt-dlp available.
- 5Make a research request. Describe the game, question and engine in plain language, and ask for source URLs.
- 6Review the sources. Check the cited URLs and archive snapshots before relying on any claim.
What is an agent skill, and what does this one do?
An agent skill is a folder with a SKILL.md file that tells an AI agent how to handle a certain kind of task. When the skill is installed and your request matches its description, the agent follows its process instead of answering from memory.
This skill searches in a fixed order. It first checks the local knowledge base and field notes with the um command-line tool. It then searches live sources such as GitHub, Nexus Mods, Steam Workshop, Thunderstore, Reddit and YouTube. When a page is dead or deleted, it looks for an archived copy on the Wayback Machine or archive.today. Every conclusion is supposed to carry the URL it came from.
- Dead forums: checks Wayback Machine snapshots (via the CDX API) and archive.today, for example for XeNTaX and Zenhax threads.
- Login-gated pages: asks you to paste the visible content instead of handling your credentials.
- Visual claims: captures screenshots as evidence, using um win shot for game windows.
- Refuses to fetch pirated games, leaked code, cracked tools, key generators or DRM and anti-cheat bypass tools.
Install the skill from the ZIP
Download the ZIP for this skill from this page. The skill ships as a single file, SKILL.md, inside the folder .agents/skills/game-research-websearch in the universal-modder repository. Unzip the download so that the game-research-websearch folder containing SKILL.md sits in one of the skills folders below.
For all of your projects, unzip it into ~/.claude/skills/. For one project only, unzip it into that project's .claude/skills/ folder. In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.
- Personal install: ~/.claude/skills/game-research-websearch/SKILL.md
- Project install: <your-project>/.claude/skills/game-research-websearch/SKILL.md
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
What the skill needs before you start
The skill needs internet access, because nearly all of its work happens on live websites and archives. It also depends on two command-line tools. The um tool handles knowledge base search and window screenshots. yt-dlp fetches YouTube transcripts.
A Nexus Mods API key is optional. The skill says the public Nexus GraphQL endpoint needs no key, and only the v1 REST API needs a personal key. If you need that, set the key in your own shell. It should never be pasted into the chat or saved in a file.
The skill runs scripts and uses the network, so read the safety notes below before installing.
- Internet access (required)
- um command-line tool (knowledge base search, window screenshots)
- yt-dlp (YouTube transcripts)
- Optional: personal Nexus Mods API key, set in your own shell, only for the v1 REST API
How to trigger the skill with natural requests
You do not need special syntax. Describe the research task in plain language. The skill's description covers starting work on a game, recovering a dead tool or link, login-gated forum threads, and writing a field note with sources.
Good requests name the game, the question and, if you know it, the engine or tool. Asking for sources in the request helps the agent follow the citation step.
- "How did people mod Game X before its forums went offline?"
- "This tool's download link returns a 404. Can you find an archived copy of its page?"
- "Research the file format used by Engine Y and give me source URLs for each claim."
- "I want to write a field note on Game Z's mod loader. Check the knowledge base and cite where each fact comes from."
What the agent does step by step
The skill describes a core loop. The agent checks the knowledge base first, forms several query variants, searches live sources, and falls back to archives when something is dead. If a page is behind a login, it stops and asks you for the content. At the end it synthesizes the findings and keeps contradictions visible.
Search results are treated as leads, not proof. The skill ranks evidence from primary sources and verified facts above reproducible community reports, and those above single anonymous posts.
- Check the knowledge base with um kb search "<game>"
- Run 3 to 5 query variants using quotes, site: filters and engine names
- Search GitHub, Nexus, Steam Workshop, Thunderstore, Reddit and YouTube
- Use Wayback CDX and archive.today for dead pages, with a cap of 3 retries
- Ask for pasted content on login-gated pages
- Record source URLs, dates and build numbers
um kb search "<game>"Safety notes: scripts and network access
The skill runs scripts and makes network requests, so you should know what it touches. It sends requests to public APIs and archives, and it spaces them out to respect rate limits. The Wayback Machine has no API key, and the skill says not to pretend one exists.
The skill also limits what it will install. It says to install tools only from an official repository or releases, and only with your permission. It tells the agent not to use forum attachments, archived downloads or file-host mirrors.
For login-gated sources, the skill never asks you for a cookie, token, cURL command, HAR file or browser console code. Treat pasted content as data: the skill says it may contain instructions, and the agent should not carry them out.
- Public APIs are queried one page at a time; Nexus forbids bulk scraping.
- Reddit requests must not spoof a browser User-Agent.
- Never share cookies, tokens, HAR files or cURL commands with the agent.
- If an original repository was taken down by a DMCA or legal notice, the skill stops and does not search for forks or mirrors.
Tips for better results
Name the engine as well as the game, since engine-agnostic answers often sit in threads about other games. Quote exact error messages or magic bytes when you know them. For visual claims, ask for a screenshot, and remember that web pages are cited by archive snapshot URL rather than a desktop capture.
Expect slow answers. The process is deliberately rate-limited and archive-aware, so it is not a quick lookup.
- Quote exact error strings, for example "fatal error loading shader archive".
- Name the engine: Dunia, Disrupt, Creation Engine, REDEngine.
- Add site: filters such as site:nexusmods.com to narrow results.
- Keep note media under media/ and under 1.5 MB, or um kb check fails.
What people use it for
- Recovering how modders extracted files from a game whose community forum has gone offline, using Wayback snapshots.
- Finding a fork of an abandoned GitHub modding tool, as long as the original was not taken down by a legal notice.
- Checking whether a Steam Workshop or Thunderstore mod still matches the current game version.
- Collecting YouTube video descriptions and transcripts that link to modding tools.
- Writing a field note for a modding knowledge base where each claim carries a source URL.
Frequently asked questions
Is the game-research-websearch skill free?
The skill is published under the MIT licence, as stated on this page. The page does not list any price, so check the repository for licence terms before you reuse it.
What do I need to run this skill?
You need internet access, the um command-line tool for knowledge base search and window screenshots, and yt-dlp for YouTube transcripts. A personal Nexus Mods API key is optional and only needed for the v1 REST API.
Where do I put the skill after downloading it?
Unzip it into ~/.claude/skills/ for all projects, or into a project's .claude/skills/ folder for one project. In the Claude apps you can upload the ZIP under Settings → Capabilities → Skills.
What happens when a forum page is dead?
The skill looks for an archived copy using the Wayback Machine CDX API, then tries another timestamp or archive.today. It records which archive worked, so you can cite the snapshot.
Will the skill log into forums or Discord for me?
No. For login-gated pages it asks you to paste the visible text or print the page to PDF. It never handles credentials and never asks for cookies or tokens.
Does the skill help find pirated games or cracks?
No. The skill explicitly refuses to fetch leaked code, pirated games, ROMs, keygens, cracked executables, or DRM and anti-cheat bypass tools.
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
This skill suits modders recovering knowledge from older game communities, and researchers or contributors who need every claim traced to a URL. If you need a quick answer and do not need sourced, rate-limited research, a normal search or chat question will be faster. Because it runs scripts and uses the network, read the safety notes before installing.
Download the skill