Skip to content

Command Center

One screen for every agent you have running

Build a personal command center that puts Claude Code, Codex and Hermes behind a single web interface on your own computer, reachable from your phone once an authenticated tunnel is in front of it.

8 modules, 24 lessons

$99One payment, yours to keep

Tax included

After this course

After this course you can open one web page, at your desk or on a train, and see every coding agent session you have running, across providers, each card carrying its provider, account profile, model and current activity. You can launch a new session into a chosen working directory, answer a permission prompt as a button while you are nowhere near the machine, stop a turn without losing the conversation, and come back after a service restart to the exact same provider conversation with its history intact. The whole thing runs on your own computer, your provider credentials stay in the provider's own local configuration, and the phone URL sits behind a Cloudflare Access policy you created before the tunnel route was ever published.

What you will be able to do

  • Launch a session with an explicit provider, account profile, model and working directory, and see every live session and every saved conversation on one screen.
  • Answer a provider permission request as a button from your phone, and watch the turn continue.
  • Resume the exact provider conversation after the web service restarts, with the saved record intact.
  • Add a second and third provider through one twenty one capability contract, so a provider quirk stays inside its adapter instead of leaking into shared interface code.
  • Publish a phone URL behind a Cloudflare Access application created before the tunnel route, then prove that an unauthenticated request and an unauthorised identity are both refused.
  • Run the 93 check release checklist twice from a cold start, with every skipped check recorded as not_applicable against the decision that made it so.

Who this is for

  • You keep three or four agent sessions alive at once and have lost track of which terminal tab holds which account.
  • You have walked away from your desk and come back to an agent that stopped eleven minutes in, waiting on one keystroke.
  • You want a phone URL into your own machine and you want the security done properly, with an identity check in front of it.
  • You are comfortable enough in a terminal to authenticate accounts, and you want a coding agent to do the engineering while you make the decisions.

Who should skip it

  • You want a finished app to install. The kit is a specification of 26 tasks that a coding agent executes with you over several focused sessions, and the folder contains no dashboard waiting for your logo.
  • You want to reach a machine you do not control, or let a teammate log in. The specification rules both out, and every hour spent arguing with it is an hour you lose.

Inside the course

Every module and every lesson, exactly as they sit in the course today.

  1. 1What You Are Building3 lessons

    Module 1

    The five local layers, the six capabilities that decide whether you open the dashboard on a Tuesday morning, and the security boundary stated before you commit to anything.

    • What a personal AI command center actually is7 min
    • The three levels, and what each one costs you in time7 min
    • The security boundary, stated before you start7 min
  2. 2The Kit and How To Hand It To An Agent3 lessons

    Module 2

    Five documents, the precedence rule that settles it when two of them disagree, and the single paste that hands ownership of the build to a coding agent.

    • The five files and the job each one does7 min
    • Handing the build to the agent with one paste7 min
    • The autonomy contract, what the agent owns and what it asks you7 min
  3. 3The Owner Interview3 lessons

    Module 3

    Ten sections of decisions: product name, host machine, providers, account profiles, remote access, login method, permissions, and the one paragraph that defines your first release.

    • Product identity, your machine, and which providers to start with8 min
    • Account profiles, remote access, and how you log in8 min
    • Permissions, idle policy, attachments, and your first release paragraph10 min
  4. 4Level One, A Dashboard That Works3 lessons

    Module 4

    Foundation before interface, the evidence rule that forbids a timer from inventing success, and the smallest slice that is genuinely useful: one provider, one session, one restart.

    • Foundation first, memory files, contracts and storage7 min
    • The evidence rule7 min
    • Your first working slice7 min
  5. 5Level Two, Multi Provider3 lessons

    Module 5

    One twenty one capability contract covering Claude Code, Codex and Hermes, plus account isolation, approval cards, attachments and exact conversation recovery.

    • One contract, every provider7 min
    • Adding Claude Code, Codex and Hermes8 min
    • Isolation, approvals, attachments and recovery7 min
  6. 6Level Three, A Reliable Daily System3 lessons

    Module 6

    Health checks, service startup, safe restart recovery, test first as the working method, and the mobile and privacy work that decides whether you still use this in month three.

    • Test first is the working method7 min
    • Health, startup and safe restart recovery7 min
    • Mobile polish, privacy and accessibility7 min
  7. 7Phone Access and the Security Boundary3 lessons

    Module 7

    Access application first and tunnel route second, credentials that never reach the browser, and the honest limit of a passing test suite before routine remote use.

    • Access to the dashboard is access to your computer8 min
    • The tunnel and the Access application, in the right order8 min
    • Credentials stay local, and the honest limit of testing8 min
  8. 8Release, and Running It Every Day3 lessons

    Module 8

    The 93 check gate run twice from cold, the four commands you write down outside the chat window, the restore drill, and the backlog that keeps the build improving.

    • The release checklist is the gate8 min
    • The four memory files and the handoff when context runs out8 min
    • Running it every day, and what to build next9 min

What comes with it

  • Every lesson in the course, with its written material and its downloads.
  • One payment. The course stays on your account and opens on any device.
  • A certificate once you finish it.
  • Five agents from the library of 81, open with any purchase on this store.

Price

Command Center

$99Lifetime access

8 modules, 24 lessons

Or take all 16 courses

$497$1,584

You save $1,087, 69% off

16 courses, 151 modules, 532 lessons

Three courses bought separately already cost more than half the bundle. The bundle carries all sixteen for one payment.

Questions

Do I get finished code, or do I build it?

You get the specification, the operating prompt for the agent, the decisions worksheet, the release checklist and the bootstrap paste. A coding agent writes the application into a new repository that belongs to you, and the course teaches you how to read what it produces, where it drifts, and how to refuse a task that only appears complete.

Is this beginner friendly?

It expects basic terminal experience and at least one authenticated coding agent account. The engineering is performed by the agent, so what you supply is the decisions, the account logins and the approvals at each phase gate. If you have never opened a terminal, take AI Foundations and Setup first and come back to this.

What software do I need, and does any of it cost extra?

Git, Node.js and at least one authenticated provider account such as Claude Code or Codex. A permanent phone URL also needs a Cloudflare account and a domain managed through Cloudflare. Those subscriptions and registrations stay yours and are billed to you, and the build resells model usage to nobody.

Why buy this one course at $99 when the bundle is $497?

One course is $99, and all sixteen bought separately come to $1,584. The bundle is $497 for the same sixteen. Buy this one on its own if the command center is the only thing you came for. The moment your cart reaches $497 the bundle is the cheaper purchase, and the catalogue offers you the swap.

Are the agents included?

Any purchase unlocks five of them: prompt-engineer, meta-ad-creative-engine, keyword-research-agent, klaviyo-flow-builder and playwriter-browser-operator. The remaining 76 stay visible and locked. You can buy one, buy the pack of 81, or join the Skool community where the whole library is included.

Can I get a refund?

These are digital goods, delivered to your account the moment payment completes, so returns are unavailable once access has started. The full module list and the outcomes are on this page for exactly that reason. Email support@murasaki.ai with any question before you buy and you will get an answer.

Start today

$99One payment, yours to keep
See the 16 courses