Flowscale
FOR AGENCY OWNERS WHO PITCH · ENGINE #1 OF THE FIELD KIT · RUNS IN YOUR OWN CLAUDE

// THE DOSSIER ENGINE · INSTALL IT, OWN IT

Stop buying the file.
Own the machine.

A CodingMash dossier costs $17 USD, built for you. The Dossier Engine is the machine that builds them, installed into your own AI. $47 USD once. Unlimited companies. No monthly fee to me, ever.

See what's inside ↓

Runs entirely in YOUR Claude account · get-you-running-or-refund guarantee · keep the files either way

// THE 3 HOURS NOBODY DOES

Every pitch you lose in the first two minutes was lost before you walked in.

Researching one prospect properly takes about 3 hours. Their site, their ads, their reviews, their competitors, who actually decides, and why they'd buy now instead of next quarter.

You have 40 prospects and a business to run. So the research becomes 10 minutes of skimming, three lazy notes, and a pitch that opens with a guess. The prospect can tell. They always can.

And the cost isn't just the lost pitch. It's the week you spent chasing a company that was never going to buy, because nobody scored the fit before the chase started.

// THE WRONG FIX

Another $99-300/mo USD data tool is not the answer.

You've seen those tools. Contact lists, intent scores, enriched fields. Data dumps. Useful, sometimes, but a list of emails is not a case. None of them tell you the thing that wins the room: what to lead with, what they'll object to, and what to say back.

That part has always been human work. The 3 hours. Which is exactly why it never gets done.

So here's the different move: instead of renting another database, install the machine that does the 3 hours and writes the conclusion.

// WHAT IT ACTUALLY IS

A real skill, installed in your own AI. Not a prompt pack.

A "skill" is a file of operating instructions your Claude reads and follows, the same mechanism real agencies run their research on. You type one sentence: "Run a prospect dossier on Cedar Park Family Orthodontics." The engine runs the 8 passes with live web search and writes the dossier into your folder.

PASS 1 · THE SURFACE

Their site: what they sell, how it converts, where it leaks.

PASS 2 · THE MONEY

Ad libraries: are they spending, and is the spend leaking?

PASS 3 · THE REPUTATION

Reviews, especially the 1-3 stars: pain in the customer's own words.

PASS 4 · THE PEOPLE

Who decides, who gatekeeps, and one human detail to open with.

PASS 5 · THE MOTION

Buying triggers: hires, locations, seasons, news. The "why now."

PASS 6 · THE BATTLEFIELD

2-3 real competitors: who's outspending and out-reviewing them.

PASS 7 · THE GAPS

Cross-reference it all into 2-4 pains an agency can actually fix.

PASS 8 · THE VERDICT

Fit score, objections with counters, the angle, 3 ready openers.

THE PART THAT MAKES IT TRUSTWORTHY

Every claim in the file is labeled [observed], [self-described], or [inferred], with a confidence level. It never invents a number. What it can't verify goes in a named gaps list, with "verify on the call" attached. A named gap reads as rigor. A confident guess reads as ChatGPT.

// THE OUTPUT, NOT A MOCKUP

This is the file it writes.

A real dossier from this exact system: Cedar Park Family Orthodontics, FIT 84/100, pursue. Fit breakdown, pain signals with evidence, buying triggers, objections with counters, the angle, three openers. The engine writes this same format into your prospects/ folder, for any company you name.

A real prospect dossier produced by this system
Real screenshot of the deliverable format. Read the full sample → the dossier

// WHAT SHIPS

Every file, annotated. No mystery box.

my-agency/ · THE DOSSIER ENGINE
├── icp.md <- your ideal client, offer, proof points (template + worked example)
├── prospects/ <- dossiers land here, one per company, forever yours
└── .claude/skills/prospect-dossier/
    ├── SKILL.md <- the engine: the full 8-pass recon procedure
    ├── fit-scoring.md <- the 100-point rubric (budget, pain, access, urgency, fit)
    └── dossier-format.md <- the case-file template, identical for company #1 and #400

Plus the install walkthrough on the delivery page, written for someone who has never opened a terminal. You get the source in copy-paste blocks AND a zip. Edit anything; it's yours.

// WHAT YOU ACTUALLY NEED · READ THIS BEFORE BUYING

The honest requirements.

Why it works this way: I never resell Claude access, and your prospect list never touches my servers. The engine runs entirely inside YOUR account. That's the design, not a disclaimer. You wanted in on AI. This is what in looks like: the real stack, in your hands.

// HOW IT WORKS

Three steps. The third one is the rest of your career.

01

Install

Unzip into a folder, install Claude Code, log in. About 5 minutes with the walkthrough.

02

Teach it your ICP

Fill icp.md: what you sell, who buys it best, your real proof points. 10 minutes, once.

03

Name a company

One sentence. Minutes later the dossier is in prospects/. Tonight, on a real prospect.

THE SYSTEM BEHIND IT

This engine is the research discipline I run my own agency on. The same discipline, pointed at my own pipeline, has pulled leads at $21 against a $35-75 USD band, inbound calls at $5-6, and 400+ leads traced to the exact ad that made them. Ask me on a call and I'll show you the ad-account receipts.

// THE MATH

$47 USD. Once.

CodingMash builds you one file for $17. The engine is $47 and then every file is free. Ten prospects in, you've paid $4.70 a file. A hundred in, $0.47. The $99-300/mo USD tools keep charging whether you pitch or not.

The guarantee: if you can't get it running, I will personally get you running, or refund the $47. And you keep the files either way. The only way you lose money here is by not pitching anyone.

// STRAIGHT ANSWERS

Questions a smart buyer asks.

I've never opened a terminal in my life.+

Then the walkthrough was written for you. It's click-by-click with screenshots: install one app, paste one command, log in with the Claude account you already have. The whole point of this engine is that the machine does the technical part. About 25 minutes, once, and most of that is filling in your ICP. If you genuinely get stuck, email me and I'll get you running, that's the guarantee.

Is this just a prompt pack?+

No. A prompt pack is text you paste and hope. This is a skill: a structured procedure your Claude loads and executes, with a scoring rubric, an output format it follows every time, and rules it isn't allowed to break (no invented numbers, labeled claims, named gaps). The difference shows up on company #30, when every file still looks identical and you can still trust the fit score.

Why not just ask ChatGPT to research a company?+

Try it. You'll get a confident page of plausible-sounding general knowledge, no fit score against YOUR offer, no objections with counters, and at least one detail it made up. The engine's edge isn't the AI, it's the procedure: 8 specific passes, claims labeled observed vs inferred, scored against the ICP you defined, in a format built for closing. The AI is the labor. The procedure is the product.

Does my prospect list touch your servers?+

No. The engine runs entirely inside your own Claude account, and the files live on your own machine (and your own GitHub, if you set that up, which I recommend). I never see who you're researching. I couldn't if I wanted to. Anthropic's terms govern your Claude account, same as they would anyway.

What does running it actually cost me?+

The Claude Pro plan, $20/mo USD, paid to Anthropic. Pro's usage limits handle a handful of dossiers a day comfortably; if you ever scale to dozens daily, Anthropic's Max plan exists. Either way you pay them, not me, and the plan does everything else Claude does for your business too. There is no per-file fee and no subscription to me, ever.

What if Anthropic changes pricing or terms?+

Then your costs change with them, exactly like your email tool, your CRM, and your ad accounts. I'll say it plainly: this product depends on a platform I don't control. What you own outright is the procedure, the rubric, and every file the engine ever writes. Those are plain text on your machine and nobody can re-price them.

So what's CodingMash for, then?+

CodingMash is for the file you want this week without installing anything: $17 USD, a real operator builds it by hand, same day usually. The engine is for owning the capability. Plenty of people do both: taste it with CodingMash, then install the machine. If you bought a CodingMash file, you've already seen exactly what this engine writes.

Will it work for my niche?+

The engine is tuned for agencies and consultants selling to SMB and mid-market businesses, anything with a website, reviews, ads, and a findable owner. Local services, healthcare, ecom, B2B services, SaaS. It's weakest on giant national brands (different decision structure, flagged in the file when it detects one). The ICP file is yours, so the scoring bends to YOUR niche, not a generic one.

// FOR THE SCROLLERS

The short version: the Dossier Engine is the prospect-research machine behind CodingMash, packaged as a skill you install in your own Claude account. Name a company, it runs an 8-pass recon with live web search, and writes a closing-ready dossier: fit score against YOUR ideal client, the live pain with evidence, buying triggers, objections with counters, the angle, and 3 openers. Every claim labeled, no invented numbers. You need Claude Pro ($20/mo USD, from Anthropic) and about 25 minutes of setup with a hand-holding walkthrough. $47 USD once, unlimited companies, keep-the-files guarantee.

// ONE ROOM OF A BIGGER HOUSE

Research is the first machine. It isn't the last.

The full system this comes from also writes the outreach, preps the meetings, drafts the proposals, and watches client health, sibling engines that share the same files. They ship one at a time in the Field Kit. The next one: The Outreach Engine, which turns any dossier in your prospects/ folder into openers and a sequence in your voice.