Content Brief Intake Before You Draft: The Ritual Creators Skip

By AgentOps Lab

You already paste ideas into an agent. What still fails is starting the draft before anyone agrees what the piece is for. A hook, a voice memo, and a pile of notes go into a blank chat; a long draft comes out; and only then do you notice the audience was assumed, the promise did two jobs, and the proof was never gathered.

This post is a content-brief-intake ritual for newsletter writers, creators, and indie marketers who already use AI agents. It is not a virality guide, not an income promise, and not an essay mill.

You will leave with a brief shape, a 15-minute drill, and a path to try one installable SKILL.md for free — on channels you already use.

The symptom: the blank-chat tax

Typical week:

1. Monday — you dump a messy idea into chat and ask for “a draft.”
2. Tuesday — the draft is aimed at a reader you never named.
3. Wednesday — a number appears that you cannot stand behind.
4. Thursday — you add a CTA the brief never supported, or you bury the one you actually wanted.
5. Friday — you rewrite from scratch because the promise and the ending disagree.

That is not “the model being dumb.” It is missing content-ops. Drafting needs a job description the agent can follow: restate the piece → name the audience (stated vs assumed) → lock one reader promise → capture format or mark it UNKNOWN → split proof mentioned vs still needed → note constraints → record a CTA only if you stated one (or NONE) → ask ≤5 questions that unblock a fair outline.

Skip that ritual and every later step — newsletter outline, thread-to-longform, critique, publish check — is guessing.

Brief shape you can paste into a skill

Working title:
Intended piece (1 sentence):
Audience (stated / assumed):
Reader promise:
Format (or UNKNOWN):
Angle (teach / story / teardown / checklist / opinion):
Proof mentioned:
Proof still needed:
Constraints:
CTA (or NONE):
Unknowns (blockers):
Clarifying questions (≤5):
Suggested next step: outline / gather sources / park

Rule: the agent must fill audience, reader promise, and proof still needed before it offers to write the piece. If those are empty, the brief is theater. If the idea is doing two jobs, park the second idea rather than stuffing both into one promise.

Five fields people skip (and why they matter)

Audience, stated vs assumed. “Developers” is not a reader. “A maintainer who already ships a small library and is tired of rewriting release notes by hand” is closer. If you did not say it, the skill should label the audience as assumed.

One reader promise. What should the reader understand, decide, or try? If the piece teaches a workflow and argues a hot take, split it. One brief, one promise.

Format, or UNKNOWN. Newsletter, blog, thread, video talking points, docs — only if you said so. The skill should not pick a platform to chase reach. An honest UNKNOWN is better than a fake “this should be a thread.”

Proof mentioned vs still needed. Examples, screenshots, quotes, numbers. Leave the “still needed” column empty of invented evidence. If a load-bearing claim has no source you can stand behind, the next step is gather sources — not draft anyway.

CTA, or NONE. Reply, subscribe, a page you already own — only if you stated it. NONE is a valid, often better, outcome. No scraped-list or cold-DM CTAs. Owned channels only.

Constraints (length, tone, embargo, “do not mention,” accessibility) sit beside these. Missing constraints stay missing; they are not filled with house style the agent invented.

15-minute drill (use a sanitized idea)

Done looks like: a brief you would actually outline from. If you would not start an outline, the skill or the input is incomplete — not “the model.”

A fair next step is one of three: outline, gather sources, or park. “Write the whole issue in one shot” is not a success condition for intake.

Hard rules worth putting in SKILL.md

• Do not write the full essay, paper, or newsletter issue at intake time (brief only)
• Do not invent audience size, open rates, testimonials, or sources
• Do not treat this as academic ghostwriting or an essay mill
• Do not draft cold DMs, scraped-list outreach, or engagement-bait hooks
• Do not pick a platform just to chase reach
• If context is missing, ask — do not fabricate the brief
• You remain the author of whatever you later draft and publish

These rules are the difference between a reusable brief and a confident, unusable draft.

What this ritual is not

It will not make a post go viral, grow a list, or raise income. Files and rituals do not promise that.

It is not a license to outsource student essays, take-home exams, or anyone else’s copyrighted thread. Brief, outline, and critique your public-facing work on channels you already use.

It is not a substitute for taste. A clear brief still needs your judgment about what is true, what is kind, and what is worth saying.

Try the free teaser

Install Content Brief Intake from the free Content Creator Ops teaser (1 skill · sample prompts · content-brief template):

→ Content Creator Ops free teaser on itch

Hub page: AgentOps Lab — Content Creator Ops

The full pack (optional, after you like the brief format) adds newsletter outlines, thread-to-longform spines, ethical repurpose maps, a capacity-honest week calendar, draft critique, and a publish checklist:

→ Content Creator Ops Pack on itch

No virality, follower, ranking, or earnings promises. Owned channels only. Not an essay mill. Not a cold-spam kit. Files only; you remain the author.

AgentOps Lab builds installable ops skills for AI agents. Deep links only — we do not ask you to trust anonymous store roots.