plugins/prompt-engineer/skills/prompt-creation/SKILL.md
Techniques and templates for creating optimal prompts following Anthropic conventions. Use when: writing a new system prompt from scratch, structuring it with Anthropic's 9-element format, or choosing emphasis levels and few-shot examples.
npx skillsauth add fusengine/agents prompt-creationInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Skill for creating high-performance prompts following 2025 best practices and Anthropic conventions.
Each system prompt should cover these 9 aspects:
# Task Context
You are an expert [ROLE] specializing in [DOMAIN].
Your goal is to [PRIMARY_OBJECTIVE].
# Tone
- Tone: [formal/casual/technical/friendly]
- Length: [concise/detailed/adaptive]
- Format: [prose/lists/tables/mixed]
# Task Description
Here are important rules:
- Always [MANDATORY_BEHAVIOR]
- Never [FORBIDDEN_BEHAVIOR]
- If [CONDITION] then [ACTION]
If unsure, say "I don't have enough information to answer."
# Examples
<example>
Input: [STANDARD_CASE_INPUT]
Output: [EXPECTED_OUTPUT]
</example>
<example>
Input: [EDGE_CASE_INPUT]
Output: [EXPECTED_OUTPUT]
</example>
# Input Data
<document>
[USER_PROVIDED_CONTENT]
</document>
# Immediate Task
Now, [SPECIFIC_ACTION] the above [document/code/data].
# Precognition
First, in <scratchpad> tags, analyze the key points.
Then provide your final answer in <answer> tags.
# Output Format
## Analysis
[Problem understanding]
## Solution
[Main response]
## Recommendations
[Additional suggestions]
# Prefill
<scratchpad>
In order of increasing effectiveness:
| Level | Syntax | Usage |
|-------|--------|-------|
| 1 | Please do X | Suggestion |
| 2 | You should do X | Recommendation |
| 3 | Always do X | Standard rule |
| 4 | YOU MUST do X | Strong rule |
| 5 | IMPORTANT: Do X | Critical rule |
| 6 | CRITICAL - ZERO TOLERANCE: Do X | Absolute rule |
1. DEFINE the objective (1 sentence)
↓
2. IDENTIFY target audience
↓
3. LIST constraints
↓
4. CHOOSE techniques (CoT? Few-shot?)
↓
5. WRITE with 9-element structure
↓
6. ADD guardrails
↓
7. TEST mentally (edge cases)
↓
8. ITERATE if necessary
testing
Copy self-audit and ban-lists — filler verbs/hype adjectives, slop placeholder names, fake-precise numbers, Title Case headlines, humor in error copy ('Oops!'), em-dash crutch, one copy register per page.
development
Logged-in web apps — dashboards, auth flows, settings, onboarding, data tables, command palettes, modals, toasts. Register `product`: density and glance-speed over marketing polish, no hero/CTA-tricks, every data surface covers empty/loading/error explicitly, tables and dataviz follow preattentive-processing rules.
development
Marketing sites, landing pages, campaign pages — register `brand` (design IS the product). Structure comes from the register's POV + a macrostructure pick, never from copying an inspiration site's section flow. Hero discipline, deviated section order, asymmetric grids, and a silhouette lookalike-test gate before ship.
development
Token-strategy core — OKLCH color rules, neutral tinting, accent-commitment levels, type scale, 8pt spacing grid, touch targets, and the canonical output format of design-system.md (the file the harness gates on). This is routing step 1 of design-method/SKILL.md — read it before design-web/design-webapp/design-ios/design-android, before picking or auditing a single color/type/spacing value.