Agent skill

create-skill

Create new Standard Operating Procedure documents following project standards. Use when documenting new processes, workflows, or guidelines. Invoked by: "new process", "create sop", "document process", "new workflow", "create documentation", "new sop", "create skill".

Stars 163
Forks 31

Install this agent skill to your Project

npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/other/create-skill-eleva-labs-agentic-coding-templ

SKILL.md

SOP Creation SOP

Version: 1.0.0 Last Updated: 2026-01-11 Status: Active


Overview

Purpose

This guide provides a standardized approach for creating new Standard Operating Procedure (SOP) documents. It ensures consistency across all process documentation, making SOPs easy to follow, maintain, and adapt into Claude Code skills.

When to Use

ALWAYS: Creating new processes, documenting existing workflows, establishing guidelines, creating templates, defining procedures SKIP: Quick notes, temporary documentation, one-off instructions


YAML Frontmatter Reference

Every skill file (SKILL.md) must begin with YAML frontmatter. This section documents all available fields based on official Claude Code documentation.

Available Fields

Field Required Description
name No Display name for the skill. Defaults to directory name. Lowercase letters, numbers, and hyphens only (max 64 characters).
description Recommended What the skill does and when to use it. Claude uses this to decide when to apply the skill.
argument-hint No Hint shown during autocomplete. Example: [issue-number] or [filename] [format].
disable-model-invocation No Set to true to prevent Claude from automatically loading this skill. Default: false.
user-invocable No Set to false to hide from the / menu. Default: true.
allowed-tools No Tools Claude can use without asking permission when this skill is active.
model No Model to use when this skill is active.
context No Set to fork to run in a forked subagent context. Only use for isolated tasks.
agent No Which subagent type to use when context: fork is set. See Agent Types below.
hooks No Hooks scoped to this skill's lifecycle.

Agent Types (for context: fork)

When using context: fork, the agent field specifies which agent runs the skill:

Built-in Agents:

Agent Description
general-purpose Default agent with full tool access. Used if agent is omitted.
Explore Read-only agent optimized for codebase exploration (Glob, Grep, Read).
Plan Planning agent for designing implementation approaches.

Custom Agents: Reference any agent defined in .claude/agents/ by path:

yaml
agent: mobile/engineer    # Uses .claude/agents/mobile/engineer/AGENT.md
agent: backend/architect  # Uses .claude/agents/backend/architect/AGENT.md

Important Notes

  • Only description is recommended; all other fields are optional
  • context: fork runs the skill in isolation WITHOUT conversation context - use sparingly and only for truly isolated tasks
  • agent only matters when context: fork is set - it has no effect on inline skills
  • Most skills should NOT have context or agent - they should run inline with full conversation context

Example: Inline Skill (Most Common)

Most skills should run inline, maintaining conversation context:

yaml
---
name: my-skill
description: >-
  What this skill does.
  Invoked by: "trigger phrase 1", "trigger phrase 2".
---

Example: Forked/Isolated Skill (Use Sparingly)

Only use forked context for tasks that truly need isolation:

yaml
---
name: heavy-analysis
description: Run deep analysis in isolation
context: fork
agent: Explore
allowed-tools: Read, Grep, Glob
---

When to Use context: fork

Use forked context when:

  • The task is completely self-contained and needs no prior context
  • You want to run a long-running analysis without blocking the main conversation
  • The task should not affect or be affected by the current conversation state

Do NOT use forked context when:

  • The skill needs to reference previous conversation messages
  • The skill modifies files that the user is actively working on
  • The skill needs to ask follow-up questions or clarifications

Quick Start

Creating a New SOP

  1. Choose Location: /docs/processes/{category}/ where category is:

    • development/ - Development workflows
    • design/ - Design and research processes
    • tests/ - Testing procedures
    • code_review/ - Review processes
    • troubleshooting/ - Diagnostic guides
    • git/ - Version control processes
  2. Name the File: Use snake_case with descriptive name

    {category}_{process_name}_process.md
    

    Examples: code_review_process.md, testing_process.md

  3. Use the Template: Copy SOP_TEMPLATE.md and fill in sections

  4. Review Checklist:

    • All placeholder text replaced
    • Metadata section complete
    • At least 2 phases defined
    • Quality checks added
    • Troubleshooting section populated
    • Related documents linked

Process Workflow

Flow Diagram

[Identify Need] --> [Choose Category & Name] --> [Copy Template]
       |
[Fill Metadata] --> [Define Phases] --> [Add Examples]
       |
[Add Troubleshooting] --> [Link Related Docs] --> [Review]
       |
[Publish] --> [Consider Skill Conversion]

Phase Summary

Phase Objective Deliverable Duration
1. Planning Identify scope and structure Outline with phases 15-30 min
2. Drafting Write core content Complete draft 1-2 hours
3. Review Verify completeness Finalized SOP 30-60 min
Total 2-4 hours

Phase 1: Planning

Objective: Define the scope, structure, and key phases of the SOP Duration: 15-30 minutes

Step 1.1: Identify the Process

  1. Answer these questions:

    • What problem does this process solve?
    • Who will use this process?
    • What is the expected outcome?
    • How often will it be used?
  2. Define the scope:

    • What is covered by this SOP?
    • What is explicitly NOT covered?
    • What prerequisites exist?

Step 1.2: Structure the Phases

  1. Break the process into 2-5 logical phases
  2. Each phase should:
    • Have a clear objective
    • Produce a deliverable
    • Have estimable duration
    • Be independently verifiable

Deliverable

Output: Process outline with phases defined

Quality Check:

  • Process has clear purpose
  • Scope is well-defined
  • 2-5 phases identified
  • Each phase has clear output

Phase 2: Drafting

Objective: Write the complete SOP content Duration: 1-2 hours

Step 2.1: Fill Metadata

markdown
# [Process Name] Process Guide

**Version**: 1.0.0
**Last Updated**: YYYY-MM-DD
**Status**: Active
**Document Type**: Process Guide
**Project**: {Project Name}

Step 2.2: Write Overview Section

Include:

  • Purpose (2-3 sentences)
  • Scope (Covered / NOT Covered)
  • When to Use (ALWAYS / SKIP)
  • Summary (2-3 paragraphs)

Step 2.3: Write Phase Sections

For each phase:

markdown
## Phase N: [Phase Name]

**Objective**: [Clear statement]
**Duration**: [Time estimate]

### Step N.1: [Step Name]
1. Action 1
2. Action 2

### Deliverable
**Output**: [What this phase produces]

**Quality Check**:
- [ ] Criterion 1
- [ ] Criterion 2

Step 2.4: Add Supporting Sections

  • Quick Reference (commands, file locations)
  • Troubleshooting table
  • Related Documentation links
  • Changelog

Deliverable

Output: Complete SOP draft

Quality Check:

  • All sections filled
  • No placeholder text remains
  • Commands are correct
  • Links are valid

Phase 3: Review

Objective: Verify completeness and quality Duration: 30-60 minutes

Step 3.1: Self-Review

Use the SOP Quality Checklist:

Structure:

  • Has YAML-like metadata header
  • Has Overview section with Purpose, Scope, When to Use
  • Has Process Workflow with diagram
  • Has 2+ Phase sections
  • Has Quick Reference section
  • Has Troubleshooting section
  • Has Related Documentation section
  • Ends with "End of Document" marker

Content:

  • Purpose is clear (answers "why")
  • Scope is defined (covered / not covered)
  • When to Use has ALWAYS and SKIP conditions
  • Each phase has objective, steps, deliverable, quality check
  • Commands are tested and correct
  • File paths are accurate

Style:

  • Uses consistent heading levels
  • Uses markdown formatting correctly
  • No spelling/grammar errors
  • Action items use imperative verbs

Step 3.2: Consider Skill Conversion

If this SOP should become a Claude Code skill:

  1. Create skill in .claude/skills/{skill-name}/
  2. Extract core workflow into SKILL.md
  3. Use progressive disclosure for details
  4. Add YAML frontmatter with triggers

Deliverable

Output: Finalized, reviewed SOP

Quality Check:

  • All checklist items pass
  • Ready for publication
  • Skill conversion assessed

SOP Quality Checklist

Required Sections

Section Required Purpose
Title + Metadata Yes Version, date, status
Overview Yes Purpose, scope, when to use
Process Workflow Yes Visual flow, phase summary
Phase 1-N Yes Step-by-step instructions
Quick Reference Recommended Commands, file locations
Troubleshooting Recommended Common issues and solutions
Related Documentation Yes Links to related SOPs
Changelog Recommended Version history

Naming Conventions

Type Convention Example
SOP File {category}_{name}_process.md code_review_process.md
Skill Folder {prefix}-{name} review-code
Skill File SKILL.md SKILL.md
Template File {NAME}_TEMPLATE.md SOP_TEMPLATE.md

Status Values

Status Meaning
Draft In development, not ready for use
Active Current, approved for use
Deprecated Being phased out, use alternative
Archived Historical reference only

Quick Reference

File Locations

Type Location
Process SOPs /docs/processes/{category}/
Setup Docs /docs/SETUP_*.md
Reference Docs /docs/*.md
Skills /.claude/skills/

Template Location

Full SOP template: SOP_TEMPLATE.md

Skill Names (New Convention)

Name Domain Location
create-skill Meta/documentation shared/
dev-feature Development workflows shared/
design-research Design/research shared/
git-pr Git operations shared/
setup-dev Environment setup {domain}/
setup-env Environment variables {domain}/
test Testing {domain}/
review-code Code review {domain}/
deploy Deployment {domain}/
help Troubleshooting {domain}/

Troubleshooting

Issue Symptoms Solution
Template not found Can't locate SOP_TEMPLATE.md Check .claude/skills/shared/create-skill/SOP_TEMPLATE.md
Unclear scope Process covers too much Break into multiple SOPs
Too detailed SOP over 500 lines Use progressive disclosure with linked files
Missing examples Users struggle to follow Add concrete examples for each phase
Outdated content Process has changed Update SOP and increment version

When to Escalate

  • Process spans multiple teams
  • Requires external tool integration
  • Conflicts with existing SOPs

Related Skills

Skill Purpose When to Use
/dev-feature Feature development workflow Creating feature documentation
/design-research Research and design process Researching before documenting

Note: Skill paths (/skill-name) work after deployment. In the template repo, skills are in domain folders.


End of SOP

Expand your agent's capabilities with these related and highly-rated skills.

Didn't find tool you were looking for?

Be as detailed as possible for better results