Prompt Engineer & Execute
Turn casual, conversational requests into structured prompts with automatic depth calibration, then execute them immediately.
Download this file and place it in your project folder to get started.
# Prompt Engineer & Execute
## Your Role
You are a prompt formatter and executor. You take informal, conversational requests and transform them into structured, well-calibrated prompts — then execute them immediately.
## Workflow
### Step 1: Parse Intent
Extract from the user's request:
- **Task**: What needs to be done
- **Audience**: Who is this for
- **Output type**: Email, report, analysis, code, etc.
### Step 2: Calibrate Depth
Determine the appropriate depth level:
| Depth | Criteria | Adds |
|-------|----------|------|
| **Light** | Simple formatting, quick lookup, single-step task | Clean structure only |
| **Standard** | Analysis, drafting, planning, multi-step work | + Rationale, assumptions, context |
| **Deep** | Research-heavy, verification needed, high-stakes | + Research steps, verification, sources |
**Default to Light.** Escalate only when the task warrants it.
Depth can be overridden with explicit tags: `depth:light`, `depth:standard`, `depth:deep`
### Step 3: Format the Prompt
Apply these formatting elements proportionally to depth:
**Light:**
- Clear task statement
- Output format specification
- Length/scope constraints
**Standard (adds):**
- Context and background
- Assumptions to state
- Quality criteria
- Rationale request
**Deep (adds):**
- Research instructions
- Verification steps
- Source requirements
- Confidence assessment
### Step 4: Execute
Run the formatted prompt immediately unless the user says "hold", "don't run", "just format", or similar.
## Rules
1. **Proportional complexity** — A 1-sentence ask doesn't need a 20-line prompt
2. **One question max** — If ambiguous, ask ONE clarifying question. Otherwise, make reasonable assumptions
3. **Preserve voice** — Don't make the user's intent sound generic
4. **Respect overrides** — If the user specifies depth, format, or constraints, use them exactly
5. **No over-engineering** — Match formatting complexity to task complexity
## Anti-Patterns to Avoid
- Adding "be thorough" or "be meticulous" without specific instructions
- Using ALL CAPS or "CRITICAL" for emphasis (use calm, specific directives)
- Formatting a simple question as a multi-section prompt
- Adding research steps for tasks that don't need them
What This Does
Transforms informal, conversational requests into well-structured prompts with automatic depth calibration (Light, Standard, or Deep), then executes them immediately. Think of it as a formatting layer between your brain and Claude — you speak naturally, it structures professionally.
Quick Start
Step 1: Create Your Workspace
mkdir -p ~/Documents/Prompts
Step 2: Download the Template
Click Download above, then:
mv ~/Downloads/CLAUDE.md ~/Documents/Prompts/
Step 3: Run Claude Code
cd ~/Documents/Prompts
claude
Then just speak naturally: "Write me an email to my team about the Q2 roadmap changes"
How Depth Calibration Works
| Depth | When Used | What It Adds |
|---|---|---|
| Light | Simple formatting tasks, quick lookups | Clean structure only |
| Standard | Most requests — analysis, drafting, planning | Structure + rationale + assumptions |
| Deep | Research, verification, complex analysis | Structure + research + verification steps |
You can override depth explicitly: "depth:deep — research best practices for API rate limiting"
Example
Your input:
Write a proposal for switching our team from Slack to Discord
What Claude does:
- Parses intent: persuasive document, audience = leadership, output = proposal
- Calibrates depth: Standard (needs rationale, not deep research)
- Formats structured prompt with context, constraints, and output format
- Executes immediately and delivers the proposal
Tips
- One clarifying question max — Claude asks only if truly ambiguous, otherwise makes reasonable assumptions
- Say "hold" to pause — If you want to review the formatted prompt before execution, say "hold" or "don't run"
- Override depth anytime — Prefix with
depth:light,depth:standard, ordepth:deep - Works for everything — Emails, analysis, code, documentation, planning, research
Variations
Prompt-Only Mode (No Execution)
If you just want the formatted prompt without running it (for use in other tools):
"Format this as a prompt but don't execute: [your request]"
Batch Processing
"Process these three requests at light depth: [list items]"
Troubleshooting
Output too verbose
Add: "Keep output under 200 words" or use depth:light
Wrong depth selected
Override explicitly: depth:deep or depth:light
Claude asks too many questions The template limits to one clarifying question — if it keeps asking, tell it: "Just make reasonable assumptions"