Content Drafter
Builds a five-line skeleton (open, stakes, body, opinion, close) and checks each line against the interview before writing. Invents nothing, traces every claim to a source, and deliberately runs a little long so the editor has material to cut.
Part of The Content Engine, a seven skill pipeline. Get every file in one place.
Where it came from
- Source: Part of The Content Engine suite, seven skills built from the one agent per task workflow Mark Fershteyn described on the GTM AI Podcast with Coach K. Published as received.
- Frameworks applied: Five-line skeleton, source tagging for every claim, voice sample matching
Why it was chosen
It separates getting the facts on the page from making the page good. Drafting and editing in one pass is where invented detail sneaks in.
Known weakness, published as found: Without an interview file or notes it will not draft. That is the point, but it surprises people expecting a topic-only prompt.
How to use it
- 1.Copy the SKILL.md text below, or download the raw file.
- 2.Create a folder named exactly content-drafter in your agent's skills directory.
- 3.Save the file inside that folder as SKILL.md.
- 4.Ask the agent one of the trigger requests below.
- 5.Check the output against what you already know before it leaves your desk.
Ask it this
- turn this interview into a first draft
- write it up from my notes
- draft this piece
Do not use it for
- make this shorter
- write five opening lines
The SKILL.md file
--- name: content-drafter description: Writes one long-form first draft using only the facts, stories, and quotes in an interview file or the user's notes, with every claim traced to a source. Use when the user says "draft this," "write a first draft," "turn my interview into a draft," "write it up from my notes," "draft from interview.md," or when the content-engine runs Step 2. Does not edit, write hooks, or format for channels. --- # Content Drafter One job: turn `interview.md` into one honest first draft. Invent nothing. A first draft should be complete and a little long. The editor cuts. The drafter's job is to get every good fact onto the page in a logical order. ## STEP 1 (REQUIRED): Load inputs 1. Read `interview.md`. If it's missing, ask the user for notes or run content-interviewer first. Do not draft from a topic alone. 2. Read files in `voice/` if present. Note: how pieces open, average sentence length, how they close. 3. Read `learnings.md` if present. Every rule there overrides this skill. ## STEP 2 (REQUIRED): Build the skeleton before writing Write a 5-line outline first: 1. **Open:** the most surprising fact or moment from the interview 2. **Stakes:** why the reader should care now 3. **Body:** the story, proof, and mistake, in the order they happened 4. **Opinion:** the take someone could argue with 5. **Close:** the one action for the named reader Check that each line maps to something in `interview.md`. If a line has no source, flag it and ask the user. ## STEP 3: Write the draft Rules: - Use the user's own words and direct quotes wherever possible. - Open on a fact or moment. No warm-up sentences like "Let's talk about..." - One idea. Anything off-topic goes in a "Next time" list at the bottom. - Receipt-grade every number: tag it in brackets with its source, like `[Q4]`. The editor removes tags later. - Target 400 to 800 words. Over is fine. Under 300 means the interview was thin: say so. ### Example **Interview detail:** "Q4: We cut SDR research from 45 minutes to 8 minutes per account using a Claude skill. 4 SDRs." **Bad draft line:** "AI dramatically improved our team's efficiency and research capabilities." **Good draft line:** "Account research used to take each of our 4 SDRs 45 minutes. Now it takes 8. [Q4]" ## Output: `draft.md` ```markdown # Draft: [working title] [Draft body with source tags like [Q4]] --- ## Source check - Claims with sources: [count] - Claims flagged with no source: [list, or "none"] ## Next time - [parked ideas] ``` ## Edge cases - **Two interview answers contradict:** use neither. List both under "Source check" and ask the user which is right. - **User asks to "make it sound smarter":** keep plain words. Smart comes from specifics. - **No reader named:** write to "a peer who does your job," and flag it.
Common questions
- What does the Content Drafter skill do?
- Builds a five-line skeleton (open, stakes, body, opinion, close) and checks each line against the interview before writing. Invents nothing, traces every claim to a source, and deliberately runs a little long so the editor has material to cut.
- Where does the Content Drafter skill come from?
- Part of The Content Engine suite, seven skills built from the one agent per task workflow Mark Fershteyn described on the GTM AI Podcast with Coach K. Published as received.. It was written by The Revenue AI Report against a 12 criterion quality rubric and graded in an independent scoring pass.
- Why was the Content Drafter skill chosen for this library?
- It separates getting the facts on the page from making the page good. Drafting and editing in one pass is where invented detail sneaks in.
- When should the Content Drafter skill not be used?
- Do not use it for: make this shorter Or: write five opening lines
- How do I install the Content Drafter SKILL.md file?
- Download the file, create a folder named exactly content-drafter inside your agent's skills directory, and save the file inside it as SKILL.md. The agent loads it when a request matches the description.
Raw file: https://www.therevenueaireport.com/agent-skills/content-drafter/SKILL.md. Plain-language skills with worked examples live in the Skills and Prompts library.
