SOP-to-Skill Converter — User Guide
NordicPulse AI tool for converting Standard Operating Procedures into structured AI agent skill files.
What is the SOP-to-Skill Converter?
The SOP-to-Skill Converter takes your existing SOPs and converts them into three structured files that AI agents can use directly:
- SKILL.md — Complete skill file with role, prerequisites, step-by-step instructions, rules and output format
- Trigger rules — Defines when and how the agent should activate the skill (automatic, manual, context-based)
- Validation checklist — Functional, quality and security checks to verify the agent follows the SOP correctly
The tool supports 6 AI tools: Windsurf Cascade, Claude Code, Cursor, GitHub Copilot, ChatGPT and a generic format.
How to use
1. Fill in the form
- SOP name (required): Give the SOP a short, descriptive name (e.g. "Customer onboarding")
- SOP content (required): Paste the full SOP text. Supports plain text and Markdown. Max 8000 characters.
- Target tool (required): Select which AI tool you will use the skill file with
- Output language: Norwegian, Swedish, Danish or English
- Department (optional): E.g. "Customer Service", "Engineering", "Sales"
- Industry (optional): E.g. "SaaS", "E-commerce", "Consulting"
2. Start the conversion
Click "Convert SOP to Skill". The tool runs three AI calls in sequence:
- Analyzes the SOP structure
- Generates SKILL.md
- Generates trigger rules
- Generates validation checklist
A progress indicator shows which phase is running.
3. Review the results
When conversion is complete, you will see:
- Metadata cards: Number of steps, triggers, checks and target tool
- File preview: Click each file to view its contents
- Copy: Copy a single file to clipboard
- Download: Download a single file or all three at once
Understanding the results
SKILL.md
The main file that defines agent behavior. Contains:
- Role: Who the agent is in the context of this skill
- Description: Brief summary of what the skill does
- Prerequisites: What needs to be in place before the skill can run
- Steps: Numbered instructions the agent follows
- Rules: Hard constraints the agent must never break
- Output format: How the result should be structured
Trigger rules
Defines three types of triggers:
- Automatic: Activates when specific conditions are met (e.g. new file detected)
- Manual: Activates with explicit command or slash-command
- Context-based: Activates based on conversation or codebase content
Validation checklist
Three categories of checks:
- Functional: Verifies each SOP step is executed correctly
- Quality: Checks that output meets quality requirements
- Security: Verifies no security requirements are violated
Cost
10 credits per conversion. You receive three files per run.
Data integrity
- All AI calls use
aiProxy.gemini()— authenticated and logged - No data is stored permanently — results exist only in your browser
- SOP content is sent to Gemini for analysis and returned as structured files
- No third parties receive your data beyond the AI model
Tips
- Detailed SOPs produce better results. The more specific the SOP, the better the skill files.
- Choose the right target tool. Each tool has different conventions for skill files (e.g. Claude Code uses CLAUDE.md).
- Use department and industry. These optional fields give the AI better context for generating relevant triggers and checks.
- Edit the files afterwards. The generated files are a starting point — customize them for your specific needs.
- Test the validation checklist. Run through the checklist items manually before deploying the skill in production.
Related tools
- Agent Security Scanner — Scan AI agent security for any domain
- AI Consultant — General AI advisory and strategy suggestions
- Compliance Generator — Generate compliance documents and policies
Last updated 2026-03-18