What is a Pack

A Pack is a recipe. Your harness executes it. That's the whole product.

You bought the harness already. Claude Code, Codex, Cursor, n8n, Zapier, a shell script your nephew wrote in a weekend. It runs on your tokens, in your account, against your data. The Pack is the standing orders the harness follows so the work happens without you sitting on it. You're the Boss. You tell the Co-pilot where to go.

Most Bosses are using AI. A few are running their business on it. The difference is whether the work happens because you typed a prompt, or because the harness had standing orders and the morning brief was in your inbox before you woke up. A Pack is what gets you to the second one.

What is a Pack?

A Pack is a structured set of standing orders, source documents, and proof artifacts the harness executes on a defined cadence. Plain English, the recipe. The harness is the kitchen. You bought the kitchen. We sell the recipes.

Each Pack maps to one bleed. Get-Paid Engine when the Case Call names that bleed, because invoices ship and money doesn't. Outbound Engine, beta $197 (was $497), because there's no outbound pipeline at all. Lead Rescue System is free because cold leads dying in your inbox is the most common bleed and we're not charging you to plug the most basic hole in the bucket.

Three things travel inside every Pack. Standing orders that tell the harness what to do and when. Voice and source assets that keep the output sounding like you instead of like a model. Receipts and watchdogs that tell you when the Pack ran, what it produced, and what it skipped. That last piece matters because most "AI workflows" are a black box where the operator can't tell whether the work happened or whether the agent quietly failed at 3am.

What a Pack isn't

A Pack is not SaaS. We don't host the work. No portal we shove you into, no usage meter ticking, no per-seat fee to let your team see the output. The work runs on your machine, in your harness, on your tokens. We sell the recipe; you cook the meal.

A Pack is not an agent. Agents are workers. A Pack is the playbook the workers run. The same Pack can be executed by ten different agents inside ten different harnesses and produce the same result. If somebody tells you their "AI agent" is the product, ask what happens when the underlying model gets deprecated. Then ask the same about a recipe written in plain language with versioned source docs.

A Pack is not automation in the Zapier sense. Zapier wires triggers to actions: when X fires, do Y. A Pack wires a bleed to a fix: where money is leaving the business, here's the standing orders that stop it. Zapier asks "what fired." A Pack asks "what stopped bleeding."

A Pack is not a prompt library. Prompts are one-shot. Packs run on a schedule, against a memory the harness keeps, with proof artifacts that compound. A prompt asks once. A Pack asks every Monday at 6am, remembers what it asked last week, and tells you when the answer changed.

The shorthand we use internally: there's a Pack for that.

Why is the harness the buyer's, not ours?

Because the harness is where your data, your IP, and your competitive advantage live. We're not asking you to upload your client list, your unreleased offer docs, your customer tickets, and your voice transcripts into a SaaS we host so we can charge per seat to read them back. Three reasons the architecture is built this way, in order of how much each one matters.

First, ownership. If we get hit by a bus, lose our funding, get acquired by a PE firm that triples the price, or stop being the right vendor for you, the recipes you've installed keep running. The harness doesn't care that we exist. It runs on the standing orders sitting in your repo. Fork them, edit them, version them, hand them to a developer to extend. That's not the deal you get with a SaaS where the workflow lives on their server.

Second, leverage. You're already paying for the harness. Claude Code Max is $200 a month. Cursor Ultra is $200 a month. n8n self-hosted is free. We're not going to charge you a second SaaS subscription on top of that to wire a few standing orders into a tool that already runs on your machine. The Cockpit runs from $49 a month. The Pack is one-time. The harness is yours.

Third, compounding. The longer you run a Pack inside your harness, the smarter it gets, because the harness keeps the memory. The Trust Pack stores your testimonial intake, your client archetypes, your authority assets. Six months in, your Pack has more useful context than the model itself. If the recipe lived on our server, that memory would be stuck behind our login. It lives on yours. Ours to design, yours to grow.

What's in the recipe?

Each Pack ships as a folder of plain markdown and YAML plus the watchdog config the harness reads. The bones are the same across the catalog.

STANDING-ORDERS.md. The instructions: what runs, when, against which inputs, with which constraints. Auditable by a human, executable by a model.

VOICE.md. Source clips, transcripts, your phrasing. So the output sounds like you, not a model fresh out of post-training.

RECEIPTS/. Every run drops a timestamped artifact: what it did, what it found, what it deferred. This is how you know the work happened. Without receipts, you're guessing.

WATCHDOGS.md. The "alert me when" rules. A Pack that runs silently for six weeks while the model drifts is worse than no Pack at all. Watchdogs are the smoke alarm.

That's it. No proprietary runtime, no vendor lock-in. The recipe is portable because it has to be.

How does a Pack run in different harnesses?

The same recipe runs in different harnesses because the recipe is written in plain language, not in vendor-specific glue code. Here's what that looks like in practice.

In Claude Code, the Pack lives in your repo's .claude/ directory. Standing orders become slash commands and skills. The morning brief fires from a scheduled hook. Receipts land in the same repo Claude is already reading.

In Codex, it lives in ~/.codex/agents/. Same standing orders, same voice files, same watchdogs. Codex calls the model differently; the recipe doesn't care.

In n8n, it becomes a folder of workflow JSON imported through the API, with markdown standing orders feeding LLM nodes as system prompts. Watchdogs run as scheduled triggers.

In Zapier (n8n's pricier cousin), it runs as a Zap with the standing orders pasted into the model step. You'll burn more tasks per month than n8n would, fine if you're already on Zapier.

In a shell script, it runs as cron jobs hitting the Anthropic or OpenAI API directly. Standing orders in a prompts/ folder, receipts in SQLite. Cheapest option, works fine if you have a developer.

The Cockpit sits above all of this. Dashboard, approval queue, audit log. The place where you see "Outbound Engine ran 14 minutes ago, queued 8 messages, 3 awaiting sign-off." The harness does the work. The Cockpit shows the work. The Pack is the recipe both follow.

The Pack catalog

Live as of today, exact prices. Free first, then standalone, then Platinum, then waitlisted.

Free (available free on packs.bossmode.ing)

• Day One Operator: morning brief, weekly wins, stale approvals
• Lead Rescue System: leads go cold, this is what nudges them

Standalone, self-install, one-time

• Super Second Brain, $197: daily AI-organized digest of yesterday's work
• Get-Paid Engine when the Case Call names that bleed: invoices ship, money doesn't
• Testimonial Harvester, Case Call-scoped: peak-delight proof moments captured, NPS and FTC consent wired
• Market Intel Radar, Case Call-scoped: buyer-side research feeding proof and authority
• Client Kickoff System, Case Call-scoped: 14-day automated handoff from signed contract to kickoff
• Content Multiplier, Case Call-scoped: one signal becomes seven authority pieces
• Lead Qualifier Engine, Case Call-scoped: fit-scores inbound leads and preps them for the diagnostic
• Outbound Engine, beta $197 (was $497): full pre-built outbound system, end to end
• LinkedIn Authority Engine, $497: voice library, post pipeline, analytics
• Webinar Funnel Engine, $497: registration through replay, no manual handoffs
• High-Ticket Close System, Case Call-scoped: every discovery session pre-prepped

Gold (self or done-for-you)

• PM Engine, $197 beta self-install (was $499) / DFY scoped on a Case Call: captures work, assigns tasks, chases progress, escalates blockers, ships evidence-based reports

Platinum

• Trust Pack, $14,997+ DFY only: 12 component Packs wired as a unified OS, 48-plus standing orders, 90 days of Boss support, addresses proof and authority and risk-reversal at the same time

Waitlisted, Q3 2026

• SEO Authority Engine: rank for the questions clients actually ask
• List Grow Engine: convert page views into email subscribers

A note on vertical Packs. If a vendor offers a Roofer Pack or any niche-industry bundle today, you're being sold a horizontal Pack with a logo glued on. Real vertical Packs sit on the roadmap. Not on the shelf, ours or anyone else's, yet.

How do you pick your first Pack?

Don't pick. Diagnose first.

Bosses pick the wrong first Pack about two-thirds of the time when they pick alone. They install the Outbound Engine when the actual bleed is "leads we already have aren't being qualified." They buy the Content Multiplier when the bleed is "buyers ghost on the close." Patching the wrong hole makes the bucket emptier, not fuller.

The Bottleneck Check at bossmode.ing/bottleneck-check is 12 questions, 4 minutes. It names your top three bleeds and the Packs that stop each one, ranked by dollar leak. Many Bosses we check are bleeding six figures a year across two or three holes. The Check tells you which costs the most this quarter, so you patch the expensive one first.

If the Check names "buyers ghost on the close" as your top bleed, no single self-install Pack fixes it. Trust is built on five assets (proof quantity, proof quality, authority, delivery transparency, risk reversal) and they have to compound. That's what the Trust Pack is for, starting at $14,997 done-for-you. The trust-gap path in the Bottleneck Check at bossmode.ing/bottleneck-check?source=trust_score_legacy scores you across the seven trust dimensions in five minutes if you want the deeper read before you spend.

One last thing. Every Pack walkthrough on this site has a "Who should not buy this Pack" section. Read it before you wire money. If the bleed isn't yours, the Pack doesn't pay back. Without trust, you're a bust, and a Pack you don't need is a tax on your token budget.