Open source, MIT licensed
Whole roles rebuilt as 59 scheduled routines in the Claude Code skill format. All eight are open source under the MIT license. Premium employees, built to the same standard, ship inside the club.
The GTM Engineer has run my club launch every weekday since August 27. The other seven are built to the same standard and I bring one online when the work calls for it. Start with the role covering the work you are furthest behind on, get it reliable, then decide whether you want a second.
Every plan, Free included, opens the repo and an install prompt built for the employees you pick.
Two kinds
The same standard, two licenses, and the line between them stated plainly.
Open source, 8 roles
The eight roles below, released under the MIT license. That license governs them, not the member license.
Premium, inside the club
Any employee, routine or kit delivered only inside the club, starting with a Data Analyst in October. Pro yearly and Lifetime get them at once; Pro monthly after 90 days.
Anything you build yourself, including anything the Agent Builder writes for you, is yours outright. Section 3 of the license is the full text.
Install
About ten minutes, and the agent does most of it.
Every plan, Free included, opens the repo and an install prompt built for the employees you pick.
Get the 8 employees freeClaude Code
ReadyNative format. Leave the kit where you extracted it and register each routine as a scheduled task pointed at its folder. Do not copy them into ~/.claude/skills: they are scheduled work, not on-demand skills.
OpenClaw
ReadyReads the same SKILL.md format. Register each routine's schedule as an OpenClaw cron job pointed at the kit's routines folder.
Hermes
ReadyPoint Hermes's recurring jobs at the kit's routines folder and mirror the cadence in SCHEDULE.md, one job per routine.
OpenCode
ReadyReads the Claude Code skill format directly. Mirror the schedule table into its recurring runs, one run per routine folder.
Grok Bot
ReadyThe closest native fit: its bots already run routines on a schedule from their own cloud computer. Create one recurring task per routine and hand it that routine's SKILL.md as the run prompt.
Codex
ReadyDrive the cadence with Codex scheduled runs, one job per routine, each handed its routine's SKILL.md as the run prompt.
Antigravity
ReadySchedule each routine with the agy job runner, pointed at the routine folder rather than registered as a global pack.
Nothing is sent to a person and nothing is spent without your click. Publishing runs only on channels you configure.
Works with the agent you already use
Claude Code
Codex
OpenCode
Grok Bot
OpenClaw
HermesThe roster
Each one is a complete routine set for a job you would otherwise hire for, contract out, or keep doing yourself.

Builds your launch board, then works it every weekday
Replaces: The growth hire who would own your entire launch
Most launches die in the gap between the plan and the week. This one closes it. On its first run it researches your business from your own public presence rather than interviewing you, writes your strategy files, builds a launch command center tailored to your offer and your ideal customer, seeds the board, and registers its own schedule. You answer almost nothing.
Then it works that board every weekday. It sweeps for buying signals and contactable people, drafts today's outreach and follow-ups into your own mailbox as drafts, fills directory and press forms and leaves them open in their tabs for you, and checks that your conversion tracking is still alive before your ad spend runs against nothing. Every morning it reconciles yesterday into one brief you read with your first coffee.
It never sends and it never spends. Those two things stay in your hands, and everything up to them is its job.

Publishes, measures, and improves your search presence every week
Replaces: A content marketer plus the SEO retainer you were quoted
Every weekday it researches the live results for a target keyword and publishes an article shaped to compete. Every week it requests indexing, reviews rankings, and rewrites the posts sitting close to page one.
Because it runs real browser sessions, it publishes to platforms with no API at all, the same way a person would. This role runs my own blogs today across more than a dozen properties.

Site health, error triage, and small changes shipped through git
Replaces: The monthly maintenance retainer and the ticket queue
Daily routines watch the sites you point it at and fix what they find within the guardrails you set. Weekly routines handle the maintenance a developer defers: dependency review, performance passes, and the backlog of small changes.
Everything ships through git with a report, so you see every change before or after it lands, your choice.

Platform-native posting every day, in your voice, on your accounts
Replaces: The social manager you keep meaning to hire
Every day it drafts platform-native posts in your voice from your actual work, into a queue you approve or let flow. Each platform gets its own version, plain text with real line breaks, budgeted against the limits that break posts in production.
A weekly routine reviews what earned attention and adjusts the drafting standards, which is the part human social managers skip.

Creative production and campaign review, with you approving spend
Replaces: The percentage-of-spend agency, for the cadence work
Daily, it produces on-brand creative variants and copy for the campaigns you run. Weekly, it reads your ad dashboards through real browser sessions, the way a person does, and prepares a change list: pause, scale, test.
Spend decisions stay with you. The role prepares and explains every change, and money moves only when you approve.

Prospecting and follow-up that never goes quiet
Replaces: The SDR you cannot justify hiring yet
Every day it sources prospects that match your criteria, qualifies them against your offer, and drafts outreach in your voice for you to approve and send. You stay the sender on every message.
Follow-up is where it earns its keep: open conversations get tracked, and the next touch is drafted at the right moment, so nothing goes quiet because you got busy.

Every customer heard, every reply drafted, churn flagged early
Replaces: The support lead role, before you can afford one
Daily routines read what customers are saying across your inbox and review surfaces, draft replies in your voice for approval, and flag the conversations that smell like churn while there is still time.
A weekly routine turns raw contact into the report a good support lead writes: what customers ask for, where they get stuck, and which product change would remove the most tickets.

Watches the whole board, including your other employees
Replaces: The operator who would run your week
Every weekday it reads what every other Employee actually did: their run logs, what finished, what failed twice, what quietly stopped running a week ago. One reconciled page tells you the state of your whole operation, which is the thing that stops being obvious the moment you are running more than two roles.
Each week it goes further. It reviews the metrics you point it at, sweeps the market surfaces you care about through real browser sessions, and produces a decision-ready brief: what changed, what it means, and the three moves worth considering, argued for and against.
It recommends and you decide, and it argues against its own top recommendation before you see it. This is the role that makes the other seven compound instead of just running.
A free account gets the walkthrough lesson, the calendar and Agent Ops Chat. Pro and Lifetime add the Masterclass, the premium library with the resale license, and the premium employees as they ship.
Pro is $349 a year or $99 a month. Lifetime is $499, paid once.