MISSION BRIEFING

GETTING STARTED

Everything you need to deploy the team and start running operations.

PREREQUISITES

Before deploying agents, you need Claude Code installed on your machine.

1

Install Claude Code

Follow the official installation guide from Anthropic. Claude Code works on macOS and Linux.

2

Verify it works

Open your terminal and run claude --version.

3

Have git installed

The installer uses git to fetch assets. Run git --version to confirm it's available.

WHAT IS AGENTS ASSEMBLE

Claude Code is Anthropic's CLI for working with Claude in your terminal and IDE. It reads files, runs commands, and follows project-specific instructions—but it's only as good as the context you give it.

Agents Assemble gives Claude Code a team of specialized agents. Each agent is an opinionated persona that orchestrates skills into end-to-end workflows—code review, debugging, feature development—delivering structured, repeatable results instead of ad-hoc prompting.

ASSET TYPES

Two types. Agents are the operators. Skills are their equipment.

AGENTS

Agents are SKILL.md files that define opinionated personas. Each agent has a specialty—Aegis for code review, Deadeye for debugging, Ironclad for feature development—and knows exactly which skills to deploy.

When you install an agent, the installer automatically resolves and installs all required skills. Agents can leverage worktrees for isolation and subagents for parallel work.

~/.claude/skills/
ironclad/ SKILL.md
aegis/ SKILL.md
deadeye/ SKILL.md

SKILLS

Skills are SKILL.md files that teach Claude specific workflows. They live in ~/.claude/skills/ and are available in every Claude Code session.

Skills define structured processes like severity-rated code reviews, TDD workflows, or planning sessions. Agents compose these skills into larger operations.

~/.claude/skills/
code-review/ SKILL.md
tdd/ SKILL.md
planning/ SKILL.md

DEPLOYMENT

One command. Pick your agents. The installer handles dependency resolution.

Mission Control
>curl -fsSL assemble.sporich.dev/install.sh | bash

Deployment Sequence

1

Cache the repository

Fetches the latest assets from GitHub into ~/.agents-assemble. Subsequent runs pull updates automatically.

2

Review the roster

The installer shows every agent and skill with their current status—new, update available, or already current. Confirm once to deploy them all.

3

Resolve dependencies and install

Assets are symlinked to ~/.claude/skills/. Agent dependencies are automatically resolved and installed.

4

Deployment summary

See exactly what was installed and where it went.

FIRST MISSION

The Director agent is always installed and acts as your single entry point. Just describe what you need—Director figures out which agent to deploy.

1

Install the team

Pick any agents you want. Director is always included automatically.

$ curl -fsSL assemble.sporich.dev/install.sh | bash
2

Open Claude Code in any project

$ cd your-project && claude
3

Invoke Director

Type /director followed by what you need in plain language. No need to memorize agent names.

you: /director review the auth module for security issues
director: Deploying Aegis — code review specialist
4

Agent runs its workflow

The deployed agent takes full control—running its end-to-end workflow with severity ratings, file references, and actionable suggestions. When done, Director offers follow-up recommendations.

FILE LOCATIONS

Where each asset type deploys on your system.

TypeLocationScope
AGENTS~/.claude/skills/<name>/SKILL.mdGlobal — all sessions
SKILLS~/.claude/skills/<name>/SKILL.mdGlobal — all sessions