Prompt Library Studio

Curated collection of ready-to-use prompt templates organized by role, task type, and domain — inspired by community best practices and production-tested patterns.

When to Use

Use this library when:

• Need ready-to-use prompt templates for common tasks
• Want role-based prompts (act as an expert in X)
• Looking for prompt inspiration and starting points
• Building a team prompt library from proven patterns

Customize when:

• Your domain has specific terminology or requirements
• Templates need adaptation for your model provider
• Performance evaluation shows room for improvement

Quick Start

Role-Based Prompts

Copy
# Technical Writer
You are a senior technical writer with 10 years of experience in developer documentation.
Write clear, concise documentation that assumes the reader is a mid-level developer.
Use active voice, short paragraphs, and include code examples for every concept.
Avoid jargon unless you define it first.

Copy
# SQL Expert
You are a database performance specialist.
When given a query, analyze it for:
1. Correctness — does it produce the expected result?
2. Performance — identify N+1 queries, missing indexes, full table scans
3. Security — check for SQL injection vulnerabilities
Always suggest an optimized version with EXPLAIN output.

Copy
# Code Translator
You translate code between programming languages while preserving:
- Logic and algorithm structure
- Idiomatic patterns of the target language
- Error handling appropriate to the target language
- Comments explaining language-specific choices

Input format: ```source_language\n{code}```
Output format: ```target_language\n{translated_code}```

Task-Specific Templates

Summarization:

Copy
Summarize the following text in {length} sentences.
Focus on: {focus_areas}
Audience: {audience_level}
Tone: {tone}

Text:
{input_text}

Data Extraction:

Copy
Extract the following fields from the text below.
Return valid JSON matching the provided schema.
If a field is not found, use null.

Fields: {field_list}
Schema: {json_schema}

Text:
{input_text}

Comparison:

Copy
Compare {item_a} and {item_b} across the following dimensions:
{comparison_criteria}

Format your response as a markdown table with columns:
| Dimension | {item_a} | {item_b} | Verdict |

Core Concepts

Template Categories

Category	Templates	Example Use Cases
Development	Code review, debug, refactor, explain	CI/CD, pair programming
Writing	Blog, email, report, documentation	Content creation
Analysis	Compare, evaluate, research, summarize	Decision support
Data	Extract, classify, transform, validate	ETL pipelines
Education	Tutor, quiz, explain, simplify	Learning platforms
Business	Strategy, planning, feedback, review	Operations

Template Variables

{role}              → Expert persona
{task}              → Specific instruction
{input_text}        → User-provided content
{output_format}     → Expected response structure
{constraints}       → Rules and limitations
{examples}          → Few-shot demonstrations
{audience_level}    → Technical level of reader
{tone}              → Formal, casual, technical
{length}            → Response length constraint

Quality Tiers

Tier	Few-Shot	Constraints	Evaluation	Use For
Draft	0	Basic	Manual review	Prototyping
Standard	2-3	Detailed	Spot check	Internal tools
Production	3-5	Comprehensive	Automated	User-facing

Configuration

Parameter	Description
template_id	Unique template identifier
category	Template category (development, writing, etc.)
variables	Required input variables
defaults	Default values for optional variables
model_compatibility	Tested model providers
quality_tier	Draft, standard, or production

Best Practices

1. Start with a template, customize with examples — add 2-3 domain-specific few-shot examples
2. Match role to task — a "security expert" prompt produces different output than a "UX designer" prompt
3. Include output format in every template — explicit format specification prevents ambiguity
4. Test templates on real data — synthetic test data doesn't surface production edge cases
5. Maintain a team library — shared templates ensure consistency across features
6. Document what works — annotate templates with performance notes and known limitations

Common Issues

Template too generic for my domain: Add domain context in the role section. Replace generic examples with domain-specific ones. Add constraints specific to your industry's requirements.

Inconsistent output format: Add an explicit complete output example. Use JSON schemas or XML tags for structure. Move format instructions to the end of the prompt.

Template works with one model but not another: Different models respond differently to the same prompts. Test across providers. Adjust instruction style — some models prefer numbered steps, others prefer prose.