The complete operator system, v2.0

Run your whole solo business on Claude Code.

Claude Code is brilliant, but it forgets everything between sessions. The Operator Pack gives it persistent memory and 15 skills that fire on their own, so it stops asking and starts executing.

See what's inside
Launch price. One payment, yours forever, free v2.x updates.
10-minute install Local-first, private No telemetry 30-day money back
claude, operator
morning, where are we
☀ operator-daily-brief ▸ fired
# reading vault + memory…
 
TOP 3 TODAY
1. ship the launch post
2. close 2 open invoices
3. review the pricing tier
 
decision-log ▸ saved → vault/
The problem

Every session starts from zero. You re-explain your business, your rules, last week's decisions. "Agentic" turns into "ask me 20 questions."

The Operator Pack is the genericized, secret-free version of a system a solo operator actually uses to run a multi-brand e-commerce and AI-agent operation on Claude Code every day.

The core

A memory that compounds, not resets.

Four layers, each holding a different kind of knowledge, so the right context loads at the right time. This is the part people pay for.

1
Always-loaded rules
Your constitution: identity, positioning, hard do's and don'ts.
every session
2
Episodic auto-capture
The session diary, recorded and compressed on its own.
automatic
3
The canonical vault
Durable truth on disk: decisions, SOPs, playbooks, logs.
on demand
4
Local memory or hub (optional)
Optional semantic search — local-first on your own machine, or a shared hub across machines at scale.
optional
▸ The optional Layer 4 runs local-first embeddings (free, no API key). The index and embeddings stay on your machine; the pack itself has no telemetry.
15 skills

They fire on intent, not on a slash command.

You describe the outcome. The right skill activates, reads your memory, and does the work like an operator.

what should I do this morning_
operator-daily-briefyour day's top 3, from your own notes
decision-logcapture a decision before you forget it
memory-capturequick note to your memory inbox
memory-searchnever re-research what you already know
inbox-triageturn captures into filed knowledge
weekly-reviewa real review from your logs
evolution-auditadopt / adapt / reject any new tool
braindump-to-promptmessy braindump into a clean prompt
sop-writerturn a process into a reusable SOP
landing-copyconversion copy in your brand voice
ad-variationsbulk on-brand ad variations
cold-outreach-drafta tight, non-spammy outreach draft
pricing-sanitypressure-test a price before you set it
competitor-watcha structured competitor teardown
release-notesa diff into clean human release notes
3 subagents included
Dispatch heavier work to a focused agent and keep your main thread clean.
  • researcher fans out, verifies, and returns a cited synthesis
  • ops takes a scoped task and ships it end to end
  • adversarial-reviewer tears your plan apart before you commit
Course, templates & Codex track included
A 5-module course teaches the whole system, and you start from filled-in templates, not a blank page.
  • A 5-module course: the amnesia problem, rules, memory, the vault, scaling & safety
  • 3-tier CLAUDE.md + an AGENTS.md Codex track — works on Codex too
  • A session-start hook recipe, doctrine frameworks, a starter vault, and MCP examples
The discipline it installs

Two laws. Everything else serves them.

01
Search before act.
Before any real task, it checks your memory for prior work. You never re-solve what is already solved.
02
Write before forget.
The moment a decision or lesson lands, it gets written to the vault, before the reply. Conversational memory is never trusted.

Built for

Solo founders, indie hackers, freelancers, and one-person agencies running real work through Claude Code, who are done re-explaining their world every session.

× Not for

People wanting a hosted SaaS or a magic make-money button. This is a system you install and operate. It rewards operators.

Claude Code Operator Pack
$$39
Launch price
4-layer memory architecture
A 5-module course
15 auto-invoking skills
3 ready subagents
3-tier CLAUDE.md + AGENTS.md Codex track
Doctrine templates + a starter vault
A session-start hook recipe
Optional local-first memory (no API key)
Commercial license, client work ok
Free updates within v2.x
Plain files, no account, no telemetry. 30-day money-back guarantee.
Questions

Good to know.

Is this a plugin or just prompts? +

Neither. It is a working system of real Claude Code skills, subagents, hooks, and a memory architecture you install as plain files. Nothing to compile, no account, no runtime.

How do I install it? +

Copy a few folders into your Claude directory, drop in the constitution template, enable one hook. About 10 minutes, and there is a step-by-step guide inside.

What does "local-first embeddings" mean? +

The optional semantic search over your memory runs on a small embedding model on your own machine — free, no API key. The index and embeddings stay on your machine; text you send to Claude Code or Codex is processed by those tools under their own terms.

Will it leak my business data anywhere? +

The pack itself ships secret-free, with no telemetry and no phone-home. Whether any data leaves your machine depends on your own tools — Claude Code/Codex, any MCP servers you wire up, and (only if you choose the server-hub option) a hosted embeddings API.

Can I use it for client work? +

Yes. The commercial license covers unlimited projects of your own and client work. You just cannot resell the pack itself.

Ship it. Then make it remember.

Stop re-explaining your world to Claude. Install the operator system and let it compound.