Prompt Template Skill

When to Use

User says:

• "create a prompt for [SDK/feature]"
• "help me write a prompt to implement [X]"
• "I need to integrate [library]"
• Shows you documentation and wants an implementation prompt

Core Purpose

Transform documentation into structured implementation prompts that:

• Extract exact technical requirements from source
• Gather user's application context
• Generate copy-paste ready prompts for any LLM

Process

Step 1: Identify Source Documentation

Ask:

1. What is the official documentation URL?
2. What specific page/section covers this?
3. What is the source code directory URL? (GitHub folder/file)
4. Any GitHub repos with examples?

Step 2: Read Documentation and Extract

From the source documentation, extract:

• Installation: Exact commands with versions
• Imports: Exact import statements
• Configuration: All options with types
• Initialization: Complete working example
• Key APIs: Core methods/functions needed

Step 3: Gather User Context

Ask about their application:

1. Framework: React/Next.js/Express/Rust/etc.?
2. Language: TypeScript/JavaScript/Rust? Version?
3. Service organization: How do they structure clients/services?
4. Environment management: .env files, config service, other?
5. Error handling: try/catch, error types, logging pattern?
6. Type system: TypeScript? Strict mode?

Step 4: Generate Structured Prompt

Use the template from resources/implementation-prompt-template.md

Fill in:

• Task overview (one sentence)
• User's tech stack context
• Technical requirements extracted from docs
• Specific implementation deliverables
• Documentation references with exact URLs

Step 5: Validate

Before delivering, check:

• Source documentation URL included (specific page, not homepage)
• Source code directory URL included (GitHub folder/file where implemented)
• Technical requirements from official docs (not assumptions)
• User context gathered (or questions asked)
• Installation commands include versions
• Initialization example is complete working code
• Deliverables are specific and actionable
• No assumptions about user's patterns

Key Principles

1. Never assume - Always ask about user's patterns
2. Extract from source - Get technical details from official docs, not memory
3. Be specific - No vague requests like "set up properly"
4. Include WHY - Context about why patterns matter from docs
5. Link precisely - Reference exact documentation pages

Example Templates

Load resources/implementation-prompt-template.md for:

• SDK Client Setup (like Grid, ZK Compression)
• API Integration (REST, GraphQL)
• Feature Implementation (new functionality)
• Migration (library upgrade, framework switch)

Integration

• Use alongside zk-compression-terminology for ZK Compression specific prompts
• Reference technical precision patterns from CLAUDE.md
• Follow progressive disclosure: only load full template when generating