Knowledge Pack

AI Copywriting Compendium

Name: AI Copywriting Compendium
Brand: agenticonsult
Price: 9 EUR
Availability: PreOrder

The anti-slop copywriting doctrine your AI agent works from. Telling a model to "write naturally" doesn't work; codified craft does: 20 named anti-patterns, 10 genres, your voice locked in a profile, and a mandatory self-check before anything reaches you.

What's included

The universal craft core (8 files): the 20-anti-pattern catalog (each with cause, cue, fix, and before→after) plus readability, voice, rhythm, truthfulness, localization, and a spoken-register core for scripts

10 genre modules: news, report, marketing, explainer, video script, audio script, email, social, UI microcopy, and technical documentation (READMEs, runbooks, release notes, PR text). Each specializes the core for its job

The brand-profile system: a fill-in worksheet that locks your voice, terminology, and honesty line, so every persuasive piece sounds like you, not like a model

Named loadouts: job → exact files to load, so the agent never carries more doctrine than the task needs

The mandatory revision gate: a rubric-driven self-check your agent runs before you see a single draft

Works with your agent harness: Claude Code, Codex, Cursor or any comparable system. Plain markdown, no code, no accounts

€9

One-time purchase · instant download · agent-facing markdown set

In accordance with § 19 UStG, no VAT is charged.

The doctrine set (~24,000 words): universal core + 10 genre modules + profiles + loadouts + exemplars
The 20-anti-pattern catalog with cue, fix, and before→after for every pattern
Agent entry files included: point your harness at the folder and it orients itself

Get notified at launch

No charge, no account needed

Everything you write with it is entirely yours
Single-buyer license: use and adapt for your business without limits
No subscription, no dependencies: plain markdown your agent reads directly

The anti-slop write loop

You + your agent

brief it withYour brand & the job

The codified write loop

Stage 1 · Set up

Lock your brand profile once

Resolve the job to a loadout

Stage 2 · Draft

Draft to the codified craft

Self-check: 20 anti-patterns

Stage 3 · Gate & ship

Revise: restore variance & voice

Ship reader-ready copy

Reads human

Articles · pages · scripts · email · UI copy

the loop this pack installs

The full craft, codified

The compendium gives your AI agent the disciplines a skilled human writer applies by instinct, each one named, checkable, and ready to run:

The anti-slop catalog

20 named anti-patterns with cause, cue, fix, and before→after, from transition-word scaffolding to the hedge stack to the em-dash monoculture readers now spot instantly. Restoration over deletion: the tell-lexicon rotates every model generation; the absences don't.

Readability & rhythm

One idea per sentence, front-loaded structure, sentence-length variance, and openings and closings that carry their disproportionate weight.

The voice system

Voice and register as orthogonal axes, a dimensional method for extracting voice from samples, and the profile that keeps every draft on-brand.

Truthfulness

Never fabricate: no invented numbers, studies, or quotes. Honest attribution, weight-of-evidence, and the rule that missing proof points get flagged, not filled.

10 genres, one spine

News, reports, marketing, explainers, video and audio scripts, email, social, UI microcopy, and technical documentation: every module an instance of the same 15-principle spine.

The revision gate

The mandatory self-check loop: draft for substance, critique against the rubric with located flaws, revise. Your agent never ships a first draft.

This depth is the product. The compendium distills the craft of natural, reader-facing writing into a doctrine your AI agent executes from directly: every rule scoped so the same move can be a craft device in one place and a defect in another, every anti-pattern shown with its worked before→after, the whole set composable from a single file to the full doctrine. It practices what it preaches: read it and you'll notice it doesn't pad, doesn't hype, and doesn't hedge. Minimal input, maximum output. You brief, your agent writes, the gate holds.

Details

The doctrine

The universal craft core

8 files · governs every genre

Anti-slop as restoration, not deletion: slop is the absence of variance, specificity, commitment, and voice; the pack teaches your agent to put the four back, which is why it keeps working as models change
The 20-anti-pattern catalog: each named pattern carries its cause, the cue to notice, the fix, and a worked before→after
Readability, rhythm, voice, and truthfulness as separate disciplines, including the rule that a persona may set tone but never facts

The specializations

The 10 genre modules

news · report · marketing · explainer · video · audio · email · social · microcopy

Each module specializes the core for its job: the summary lede for news, awareness-stage diagnosis for sales pages, the muted-autoplay hook for video scripts, one canonical status verb per state for UI copy
Named loadouts wire it together: the agent loads exactly the files a job needs, from one file for a focused fix to the full set for open-ended work
A spoken-register core loads only for video and audio: written craft and spoken craft are not the same discipline

Your voice, locked

The brand-profile system

fill-in worksheet · voice archetypes · honesty & language overlays

One structured interview locks your voice axes, terminology, and honesty stance, or your agent interviews you against the worksheet and fills it in
Four worked voice archetypes to start from, plus a dimensional method for extracting any voice from writing samples
An exemplar library that grows from your own revisions: the agent saves its strongest before→after pairs, so calibration compounds

How it works

You point your agent at the pack's folder; the included agent entry file orients it automatically. The spine teaches the loadout model: a universal core every genre inherits, one thin module per genre, and profiles stacked on top. You fill the brand worksheet once, or have the agent interview you against it, and from then on every persuasive piece carries your voice, your terminology, and your honesty line. Before anything reaches you, the agent runs the mandatory self-check: draft for substance, critique against the rubric with specific located flaws, revise. Two passes minimum. That gate is what separates this from a first-draft generator.

Who this is for

Operators, founders, and small teams who produce reader-facing content with AI agents (articles, reports, landing pages, scripts, newsletters, social, product UI copy) and are done with output that reads like a model wrote it. No copywriting background required: the doctrine ships in the pack, and the included prompt blueprints make the first session productive before you've learned anything by heart. If you already run Claude Code, Codex, Cursor, or any comparable harness, this drops in as a folder.

What you need

Your own AI agent harness that can read files: Claude Code (where the agent entry loads automatically), Codex, Cursor, a Claude project, or any comparable system. Nothing else. The pack is plain markdown: no code, no dependencies, no accounts, nothing to buy from us beyond the pack itself.

License

Single-buyer license: use and adapt for your own business without limits; reselling the materials themselves is not permitted. Everything you write with it is entirely yours.

License: Single-buyer license · everything you write with it is entirely yours

€9