An AI system is not a pile of tools or a folder of saved prompts, it is a small set of files that make Ultron work like your team, and you can stand up the whole thing today with just two of them.

This is the exact base for building AI systems with Ultron, turned into copy-paste files you can steal today. No coding. You only need two files to start, everything else is an upgrade you add when you hit a real reason to.

What an AI system actually is

An AI system is not a pile of tools or a folder of saved prompts. It is a small set of files that make Ultron work like your team: it knows who you are, it runs your repeatable jobs the same way every time, and it can plug into your real accounts.

Ultron ships five building blocks for this. You do not set them all up at once. You start with two, and you add the others only when a specific thing starts annoying you. Here is the whole map:

ULTRON.md. Your always-on rules and facts. Ultron reads it in full every session. Add it day one (required).
A skill. A repeatable job Ultron runs on command. Add it day one (required).
Auto memory. Ultron writes its own notes from your corrections so it stops repeating mistakes. Automatic, you do nothing.
Subagents. A helper with its own memory that does a heavy side job and hands back a clean summary. Add it when a big task floods your chat.
Rules. Extra rule files, so your ULTRON.md stays short. Add them when ULTRON.md passes ~200 lines.
Enforcement (settings). Turns a rule into a hard block Ultron cannot cross. Add it when a rule must hold every time.
Connectors. Plugs Ultron into your real accounts (email, calendar, docs, a database). Add them when you keep copying from a tab Ultron can't see.

The trap most people fall into is trying to build all seven at once. Don't. Two files is a real, working system. The rest earns its place later.

The 2-file core (this is the whole system to start)

A real AI system is two files:

ULTRON.md, your always-on rules and facts.
One skill, a job Ultron runs on command.

Set up these two and you have a system. Everything else is optional. Here are both, ready to fill in.

File 1: ULTRON.md

Put this at the root of your project as ULTRON.md (or inside a folder as .ultron/ULTRON.md). Ultron loads it in full at the start of every chat, so it is where your voice, rules, and "always do X, never do Y" live.

Prompt

# [Your project or brand] ## What this is One or two sentences on what you're building and who it's for. ## How I work - Talk to me in plain language. Infer what I mean, act, then summarize. - Keep outputs copy-paste ready. No em dashes. - Ask before anything that deletes files or spends money. ## My voice and rules - Voice: [e.g. warm, direct, lowercase captions] - Always: [your non-negotiable, e.g. name real tools, never a vague category] - Never: [e.g. post anything without showing me the final version first] ## Where things live - [e.g. drafts in /drafts, final assets in /assets, scripts in /scripts] ## My tools and accounts - [e.g. email, calendar, docs, scheduler]

Keep ULTRON.md under about 200 lines. When a section turns into a step-by-step process, move it into a skill (file 2). Long ULTRON.md files get followed less reliably, so treat it like a table of contents, not a manual.

File 2: your first skill

A skill is one repeatable job, saved so Ultron runs it the same way every time. Make a folder .ultron/skills/your-skill/ and put a file called SKILL.md inside. The description line is what tells Ultron when to use it. Type /your-skill or just ask, and it loads only when needed, so it costs almost nothing until you use it.

Prompt

--- description: Turns a rough idea into a finished [thing] in my style. Use when I say "draft a [thing]", paste a [source], or ask to [do the task]. Do not [the one thing it must never do]. --- ## Inputs What I'll give you (a link, a topic, a rough note), and what to ask for if it's missing. ## Steps 1. [First concrete step] 2. [Second step, with the real rule or constraint] 3. [Output format: exactly what the finished thing looks like] ## Rules - [Voice and format, e.g. sentence case, no em dashes] - [Quality bar, e.g. name at least 3 real tools, cite sources] ## Output Show me the finished result inline. Save a file only if I say so.

Everything is optional in that frontmatter except the description, and even the skill name defaults to the folder name. A skill is just your plain-English playbook. If a step ever needs real code, the skill can call a small script that lives right next to it in a scripts/ folder, so you never juggle "workflows" and "scripts" as separate things.

Your first 30 minutes

Open Ultron and run /init. It drafts a starter ULTRON.md by reading your project. Treat that draft as a rough first pass.
Paste the ULTRON.md template above over it and fill in the blanks with your real voice, rules, and tools.
Pick the one task you re-explain to Ultron the most (writing a caption, building a brief, tidying notes). Write it once as a skill using the SKILL.md template.
Run it. Type /your-skill and watch it do the job your way.
The moment Ultron gets something wrong, add one line to ULTRON.md so it never repeats. That single habit is what turns a setup into a system that gets smarter every week.

Add these later (only when you feel the exact pain)

Do not build these on day one. Each one solves a specific problem, and the trigger tells you when it is time.

Auto memory

You do not build this. Ultron writes its own notes from your corrections and preferences, and pulls them back automatically. It is on by default. You write ULTRON.md, Ultron writes this. Just know it exists so you understand the two-part memory.

Subagents (.ultron/agents/)

Trigger: a heavy side job (deep research, reviewing lots of files) is flooding your main chat with output you won't reread. A subagent runs it in its own separate memory and hands back a clean summary, so your main conversation stays tidy.

Rules (.ultron/rules/)

Trigger: your ULTRON.md creeps past ~200 lines. Move a chunk (say, all your caption rules) into .ultron/rules/captions.md so the main file stays lean.

Enforcement (.ultron/settings.json)

Trigger: you want a rule to hold every single time, not just be politely requested. A line in ULTRON.md is guidance Ultron can slip on. A deny rule here is a hard block it cannot cross. This is how you protect the scary stuff:

Prompt

{ "permissions": { "deny": ["Read(./.env)", "Read(./.env.*)", "Read(./secrets/**)"] } }

Bash already asks before it runs anything by default, so you only add rules here to hard-block something or to safely loosen it. Do not bother allow-listing read-only actions, they never prompt anyway.

Connectors

Trigger: you keep copying data out of a tab Ultron can't see. Connectors plug Ultron straight into your real accounts (email, calendar, docs, a database) so it reads and acts on the real thing. In Ultron you add one through the connectors flow and check it in the connectors panel. Sign in through the secure connect flow, never paste a token into a file.

Where this goes next

This starter pack is the foundation: the two files that turn Ultron from a chat window into a system. From these two files you can grow a full system that runs your content, your income, and the back end of your business, built live, step by step, with everything unlocking on day one.