🤖

Build Your Own AI Agent Army

Set up a team of 6 AI agents that build code in parallel — with memory, checkpoints, and the Matrix CLI.

60 min

total

lessons

free

always

You've been using AI as a single assistant. What if you had a TEAM? One writes the plan, two build code in parallel, three review for bugs, security, and quality — all coordinated through a shared memory file. This is the Matrix system, and you're setting it up today.

who it's for: Developers, technical builders, anyone comfortable with the terminal and ready to level up from single-agent to multi-agent.

what you leave with: A working Matrix setup in your project — 6 agent skills, a pipeline orchestrator, memory system, checkpoints for resume, and the global matrix CLI command.

lessons

The 11 Primitives

10 min

Every AI agent system needs 11 pieces. Record, Memory, Orchestrator, Execution Env, Agent Runtime, Integrations, Quality Gates, Delivery Target, Observability, Skills, Identity. The tools don't matter — the pieces do.

activity

Review the 11 Primitives table. Identify which ones your current project already has.

The Memory System

10 min

6 agents, none remember anything. One append-only markdown file is the ENTIRE coordination layer. Three rules: never overwrite, read all / write your section, no chitchat.

activity

Read a sample memory file. Understand how agents 'talk' through the file.

Set Up Matrix in Your Project

15 min

Run matrix --init. Get 6 skill files, memory directory, .matrix config — all set up and ready to use.

activity

Initialize Matrix in your project. Open a SKILL.md and customize one rule for your stack.

Run Your First Agent Task

15 min

Start a task, run the content agent, resume into parallel dev agents, see the review. Experience the full pipeline.

activity

Run matrix --new 'Add a footer'. Execute each agent prompt. See the pipeline work.

Customize Your Skills + Source Code Tour

10 min

Teach your agents YOUR project's rules. See the orchestrator source code — one function runs all agents, swap the skill = different agent.

activity

Customize 2 skill files. Browse the kickoff-agents.ts source. Understand the agent runner pattern.

bonus: your starter prompt

paste this into ChatGPT or Claude and fill in the blanks.

You are an AI systems architect. I want to set up a multi-agent development team in my project at [PATH]. Stack: [STACK]. Create 6 SKILL.md files (content, frontend, backend, review, security, quality) with strict file ownership boundaries. Include a memory system (append-only markdown), checkpoint system for resume, and file locking for parallel agents. Follow the 11 Primitives framework. Make it work with: npx tsx scripts/matrix.ts

Build a Family Command Center App Version Control & GitHub

made with ❤️ by the GAINS community. free forever.

get the printable cheat sheet →

lessons

The 11 Primitives

10 min

activity

Review the 11 Primitives table. Identify which ones your current project already has.

The Memory System

10 min

6 agents, none remember anything. One append-only markdown file is the ENTIRE coordination layer. Three rules: never overwrite, read all / write your section, no chitchat.

activity

Read a sample memory file. Understand how agents 'talk' through the file.

Set Up Matrix in Your Project

15 min

Run matrix --init. Get 6 skill files, memory directory, .matrix config — all set up and ready to use.

activity

Initialize Matrix in your project. Open a SKILL.md and customize one rule for your stack.

Run Your First Agent Task

15 min

Start a task, run the content agent, resume into parallel dev agents, see the review. Experience the full pipeline.

activity

Run matrix --new 'Add a footer'. Execute each agent prompt. See the pipeline work.

Customize Your Skills + Source Code Tour

10 min

Teach your agents YOUR project's rules. See the orchestrator source code — one function runs all agents, swap the skill = different agent.

activity

Customize 2 skill files. Browse the kickoff-agents.ts source. Understand the agent runner pattern.

bonus: your starter prompt

paste this into ChatGPT or Claude and fill in the blanks.

You are an AI systems architect. I want to set up a multi-agent development team in my project at [PATH]. Stack: [STACK]. Create 6 SKILL.md files (content, frontend, backend, review, security, quality) with strict file ownership boundaries. Include a memory system (append-only markdown), checkpoint system for resume, and file locking for parallel agents. Follow the 11 Primitives framework. Make it work with: npx tsx scripts/matrix.ts

made with ❤️ by the GAINS community. free forever.

get the printable cheat sheet →