Copywriting
Every's general copywriting skill. Writes original marketing copy grounded in whichever brand-voice and positioning are loaded dynamically at invocation. This is the workhorse craft skill — channel-specific skills (launch-email, linkedin-post, etc.) handle format, but copywriting handles the fundamentals of writing well for brands.
The principles here synthesize the creative canon (Ogilvy, Bernbach, Wieden, and others from marketing-os) into operational writing guidance. The skill adapts to format and channel but the core craft is consistent.
System
Copywriting is composed of seven reference files:
references/principles.md
The copywriting principles we hold — synthesized from the creative canon in marketing-os. The non-negotiable standards for how Every writes.
references/formats.md
Format guidance — long vs. short, headline vs. body, manifesto vs. product copy. How format shapes the writing, and how to choose the right format for the job.
references/channel-adaptation.md
How copy changes by channel. The same message written for email, LinkedIn, X, web, and press operates differently. This reference governs those adaptations.
references/antipatterns.md
What not to do. AI tropes, cliché, LinkedIn-voice, hollow superlatives, and the specific failure modes that make marketing copy generic or untrustworthy.
references/messaging-doc-discipline.md
The writing rules for brand and messaging documents (generalizable across all products):
the four-bucket protocol, research-before-drafting, proof-travels-with-the-claim,
hypothesis tagging, full citations, the competitive-read expiry rule, and the
differentiation ladder. The verification discipline that every strategy or messaging draft
runs before delivery. Cross-references marketing-science/research/references/ (competitive
audits, synthesis methodology) rather than restating them.
references/copywriting-line-rules.md
Line-level copywriting rules (C1–C10) from the sample-messaging passes and the creative canon: no imitation compression, every pronoun needs a visible antecedent, the line must survive its own logic, generate for a kill rate, state the claim plain before styling it, match the promise to the artifact, keep brand promises out of compliance-copy register, write for the ear, make every rhetorical device earn its place, and give each copy unit one dominant job.
references/failure-ledger.md
The failure ledger—the recurring ways a draft breaks when operational facts are left unsupported or brand promises are weakened by the wrong verification standard, with the claim-classification test for any draft.
When to invoke
- When drafting net-new copy for any brand or product
- When a channel-specific skill isn't the right fit (multi-format, unusual deliverable)
- When writing copy that spans multiple channels or formats
- When a draft needs to be written from scratch rather than edited
- When the task is "write this" rather than "fix this" (editing) or "name this" (naming)
For a sequence of related assets that must learn from human edits and prove that the
learning transfers, route to craft/compound-copywriting. It orchestrates this skill,
editing, run scoring, and the Marketing OS learning loop.
Reference routing
| Task type | Load |
|---|---|
| Core writing standards and principles | references/principles.md |
| Choosing or adapting to a format | references/formats.md |
| Adapting copy across channels | references/channel-adaptation.md |
| Reviewing for failure modes | references/antipatterns.md |
| Writing or auditing a brand/messaging doc for verification discipline | references/messaging-doc-discipline.md |
| Drafting or QC-ing original brand copy at the line level | references/copywriting-line-rules.md |
| Diagnosing why a draft reads unsourced, generic, or off | references/failure-ledger.md |
Dependencies
foundation/marketing-osbrand-voice/{relevant brand}(loaded dynamically)positioning/{relevant brand}(loaded dynamically)
Status
Core principles, line rules, messaging discipline, and product-example guidance are active. Channel adaptation and most format-specific guidance remain scaffolding until their references are populated.