What is a Loadout?
A Loadout is a “ready-to-use bundle of AI tools for a task” — the MCP, Skills, Rules, Prompts and Agents you need, gathered into one set. The word is borrowed from games: the kit of weapons and gear you arrange before a mission. In short, instead of hunting down and configuring each piece yourself, you grab the Loadout that fits the job and go.
What a Loadout really is
The word Loadout is borrowed from games, where it means “the kit you prepare before a mission” — picking the weapons, gear and items that suit the level ahead. In the AI world it's the same idea: a ready-to-use set of AI tools assembled for one specific job.
Instead of finding this MCP, that Skill, another Rule, then wiring each one up yourself, a Loadout gathers everything the job needs into a single set — drop it into your AI tool and start working right away. And it doesn't need every type — some Loadouts are just an MCP and a few Rules, depending on what the task actually calls for.
What a Loadout is made of
A Loadout organizes its “resources” into 3 groups so it's clear what each part does:
🧠 Commands & abilities
- Prompt — instructions or a template that set the AI's role for instant results (read the Prompt guide)
- Rule — standards the AI must follow every time (read the Rules guide)
- Skill — specialized know-how the AI picks up on its own when a task matches (read the Skills guide)
- Agent — a sub-agent with its own context and tools, sent off to do heavy work and report back · advanced, mainly on Claude Code (read the Agents guide)
🔌 Connectors
- MCP — connects the AI to external tools and data: files, databases, the web (read the MCP guide)
📎 Files
- Knowledge — reference files you attach for the AI (PDFs, contracts, docs)
- Template — a template the AI fills in for you (PR, memo, email)
Beyond the components, a Loadout also states its Prerequisites (things to have ready first — accounts, API keys), its Workflow (the AI's step order) and Metadata (version/license) — all in one set.
Why a Loadout helps
The appeal of a Loadout is that it cuts out the fiddly setup before you start, so you:
- Start fast — no hunting down and configuring each piece; grab the set that fits and go
- Don't miss anything important — the set comes complete for the job, so you won't forget a needed Rule or MCP
- Get pieces that fit together — the components are already chosen to work as one, so you're not guessing what pairs with what
- Reuse and share — a whole team on the same Loadout gets the same working standard
In short, a Loadout turns AI setup from “assemble it yourself, piece by piece” into “pick the right set and get to work.”
Loadout examples by task
Loadouts are usually organized by type of work, so you can pick one that fits exactly — for example:
- Web-dev Loadout — MCP for GitHub + Filesystem, a code-review Skill, a coding-standards Rule, and a bug-hunting Agent, all in one set
- Content Loadout — MCP for Notion/Google Docs, a writing-and-editing Skill, headline/caption Prompt templates, and a brand-tone Rule
- Online-commerce Loadout — MCP to pull product/price data, an ad-copy Skill, customer-chat reply Prompts, and a Rule for required/forbidden wording
See curated, pre-assembled Loadouts organized by type of work in the Loadout library.
Which AIs work with a Loadout
A Loadout is a portable Markdown file, so its core can run in an AI that accepts files or long text. Advanced capabilities such as MCP, Skills and visual output depend on the client and account:
- ChatGPT, Claude and Gemini — attach the file or paste its Markdown and run the workflow in chat
- Claude Code — suited to coding Loadouts that read files, run commands or connect MCP tools
- Other AI clients — can run the prompt, rule, knowledge, template and workflow core when the full context fits
How to start using a Loadout
Getting started with a Loadout takes 3 steps:
- Pick a Loadout that fits the job — open the Loadout library and choose a set by type of work, such as web dev, content, or commerce.
- Open it in your AI — attach the
.loadout.mdfile or paste the complete Markdown, then ask the AI to load it and collect missing inputs first. - Review permissions and run — read the prerequisites and approval gates, especially before connecting MCP tools, writing files or changing live data.
For exact starter instructions, see How to use a Loadout with your AI.
What to know before using a Loadout
Loadouts are convenient, but because they bundle several pieces together, it's worth a look before you use one:
- Check what's in the set — especially any MCP that reaches real data; know what permissions you're granting
- Choose trustworthy sources — use Loadouts built from official or well-reviewed open-source components you can inspect
- Start with only what's needed — you don't have to enable every piece at once; take just what the job needs, then add more
Loadout vs a single component
The main difference is that a Loadout is a “set” combining several pieces, while a single component (MCP, Skill, Rule, and so on) is “one piece” you assemble yourself:
| Aspect | Loadout | Single component |
|---|---|---|
| What it is | A set bundling MCP/Skills/Rules/Prompts/Agents for a job | A single component that does one thing |
| Ease of starting | Grab the set and go — no assembling it yourself | You pick and configure each piece to fit together |
| Best for | People who want to start fast, or one standard for a whole team | People who want fine control, choosing each piece |
Read on: What is MCP · What are AI Skills · What are AI Rules
Build your own Loadout
If no ready-made Loadout fits your work exactly, you can build one — pick the MCP, Skills, Rules, Prompts and Agents your job needs, then bundle them into one set to reuse or share with your team.
Start with the Loadout builder — choose the components you want and it assembles a ready-to-use set for you.