Markdown Prompt Templates for Operators

The Markdown Prompt Template

Copy this template and fill in the sections for any task. Every field matters. Operators who skip sections get generic output -- operators who fill them in precisely get client-ready deliverables on the first pass.

# [Task Title -- specific enough that you could hand it to a colleague]

## Role

You are a [specific role with domain context]. You have experience with [relevant domain,
industry, or audience]. Your output should read as if it came from a [seniority level]
practitioner, not a generalist.

## Background

[2-4 sentences of context. Who is the client? What's the business situation?
What decisions will this output inform? What does the reader of this deliverable
already know?]

## Task

[One clear sentence stating exactly what the model should produce. If you need
multiple outputs, break them into numbered sub-tasks here.]

## Requirements

- [Specific constraint or inclusion #1]
- [Specific constraint or inclusion #2]
- [Tone or voice requirement]
- [Length or format requirement]
- [Anything the output must NOT include]

## Output Format

### Document Structure

[Describe the sections the output should contain, in order.]

### Formatting Elements

[Specify headers, bullet points, tables, bold text, code blocks -- whatever
the deliverable calls for.]

### Expected Sections

[List each section by name with a one-sentence description of what it should contain.]

The heading hierarchy here is deliberate. The ## sections signal major context shifts to the model -- role, background, task, requirements, and format are distinct categories of instruction, not a continuous paragraph. The ### subsections under Output Format let you get granular about structure without cluttering the top-level prompt.

One rule: the Task section gets one sentence. If you find yourself writing two sentences, you have two tasks. Split them.

---

Three Prompt Patterns for Operators

These are ready-to-deploy templates for work that comes up repeatedly across client engagements. Adapt the bracketed fields to your specific situation.

Pattern 1: Client Report Summarization

# Q3 Performance Summary for [Client Name] Board Presentation

## Role

You are a senior fractional CFO preparing an executive summary for a board audience.
You understand how to translate operational data into strategic narrative without
losing precision.

## Background

[Client] is a [industry] company with [revenue range] in annual revenue. The attached
report covers Q3 operational and financial performance. The board meets quarterly and
expects a concise summary with variance commentary -- they do not want raw data tables
without interpretation.

## Task

Summarize the Q3 performance report into a 400-word executive brief suitable for
the opening page of a board deck.

## Requirements

- Lead with the single most significant performance story, positive or negative
- Include three supporting data points with prior-period comparisons
- Flag one risk or watch item for Q4
- Write in present tense, active voice
- No jargon specific to the operations team -- this is a board-level document

## Output Format

### Document Structure

Three paragraphs: headline summary, supporting detail, forward look.

### Formatting Elements

Bold the key metrics on first mention. No bullet points -- this reads as prose.

### Expected Sections

- Performance Headline: the lead story in two sentences
- Supporting Detail: three data points with context
- Forward Look: Q4 risk or opportunity to watch

Pattern 2: Content Improvement and Editing

# Editorial Pass -- [Document Title]

## Role

You are a senior editor with experience in B2B professional services content.
You know how to tighten writing without losing the author's voice.

## Background

This document was drafted by [author/team] for [audience -- clients, prospects,
internal stakeholders]. The tone should be [professional, conversational, authoritative].
The primary goal is [inform, persuade, instruct].

## Task

Edit the document for clarity, concision, and flow. Do not rewrite the substance --
preserve all factual claims and structural decisions the author made.

## Requirements

- Cut passive voice where active voice is clearer
- Flag any sentence over 30 words and offer a shorter alternative
- Remove filler phrases ("it is important to note," "in order to," "at this point in time")
- Do not change technical terminology or client-specific language
- Return the edited document with inline comments for any change that alters meaning

## Output Format

### Document Structure

Return the full edited document followed by a change log.

### Formatting Elements

Use inline [COMMENT: reason] notation for substantive edits.

### Expected Sections

- Edited Document: full text with tracked changes noted inline
- Change Log: bulleted list of the three most significant edits and why each improves the document

Pattern 3: Meeting Prep Brief

# Pre-Meeting Brief -- [Meeting Name] with [Client/Stakeholder]

## Role

You are a chief of staff preparing a senior fractional leader for a client meeting.
You know that executives need context, not data dumps.

## Background

[Name] is meeting with [client contact and title] at [company] on [date]. The purpose
is [strategic review / quarterly check-in / issue resolution / new engagement scoping].
The relationship is [new / established / strained]. Key history: [2 sentences max].

## Task

Produce a one-page pre-meeting brief the fractional leader can read in five minutes
before the call.

## Requirements

- Open with the one thing the leader must accomplish in this meeting
- Include no more than three background points -- the most relevant context only
- List two to three questions to ask the client
- Note any landmines or sensitivities to avoid
- Keep the entire brief under 400 words

## Output Format

### Document Structure

Four sections: Objective, Context, Questions to Ask, Watch-Outs.

### Formatting Elements

Bold the section headers. Use bullet points throughout -- no prose paragraphs.

### Expected Sections

- Objective: one sentence stating the meeting's success condition
- Context: two to three bullets on relevant history or current state
- Questions to Ask: two to three specific questions, not generic ones
- Watch-Outs: one to two sensitivities or risks for this specific conversation

---

Common Formatting Mistakes

Most operators who get poor output are making one of four structural errors. The model is not failing -- the prompt is.

Heading hierarchy errors. The most frequent mistake is opening a prompt with a single # H1 for every section. When every section is H1, the model treats the entire prompt as one undifferentiated block. The hierarchy # for the title, ## for major sections, and ### for subsections mirrors the document structure the model was trained on. Break that hierarchy and you break the signal.

Missing code fences. If your Output Format section describes a code block, JSON structure, or table schema you want returned, put an example inside triple backticks (the ``` characters that open and close a code block). Describing a structure in prose is less precise than showing it. The model will match a fenced example far more reliably than a written description.

Inconsistent list formatting. Do not mix bullet points and numbered lists within the same Requirements section unless the order explicitly matters. Numbered lists imply sequence. Bullet lists imply parallel options. Mixing them creates ambiguity about whether items are ranked, sequential, or equivalent.

Walls of text in the Background section. Background should be 2-4 sentences. Operators who write 10-sentence Background sections are burying the Task. The model weighs the Background heavily -- give it more material than the Task, and the output will reflect that imbalance. Keep Background tight. Put detail where it belongs: in Requirements and Output Format.

> Hard rule: If your prompt takes longer than 90 seconds to read aloud, it is too long. Cut Background first. Cut Requirements last.

---

Building Your Prompt Library

A prompt you write once and forget is a missed asset. Operators managing three or more client engagements should treat their markdown prompts the same way they treat contract templates -- version-controlled, organized by function, and refined after every use.

The folder structure that works. Keep a single prompts folder organized by deliverable type, not by client. Client-specific context goes inside the prompt as Background -- the template itself stays generic and reusable.

/prompts
  /reporting
    board-summary.md
    weekly-ops-brief.md
  /content
    editorial-pass.md
    linkedin-draft.md
  /meeting-prep
    client-check-in.md
    new-engagement-scoping.md
  /analysis
    competitive-review.md
    market-sizing.md

Refine after every use. When a prompt returns something off -- wrong tone, wrong length, missing a section -- fix the template immediately, not the output. Five refinement passes over a single engagement produces a template that rarely needs touching again.

Capture the Requirements section carefully. The Requirements bullets are where most of the reusable value lives. A board-summary prompt that worked for a SaaS client probably has 80% overlap with what you need for a professional services client. Copy the template, update the Background, and adjust two or three Requirements bullets. That is a two-minute task, not a twenty-minute one.

What to track per template. A simple comment block at the top of each file saves time across engagements.

<!--
  Template: Board Summary
  Last used: [date] for [client type]
  Works well for: quarterly reviews, board decks, investor updates
  Adjust for: technical audiences (add jargon note to Requirements)
-->

A library of fifteen well-maintained prompts covers the core deliverable types most fractionals produce: board summaries, weekly ops briefs, meeting prep, editorial passes, competitive reviews. Building it takes one full engagement. Deploying it saves time on every engagement after that.