Core Utilities Plugin¶

Comprehensive development utilities for plugin/command/agent development, validation, CI automation, text humanization, and PDF-to-Markdown conversion.

Version 3.1.3¶

New: Added pdf-to-markdown skill for converting PDF documents to Markdown with dual-mode support (fast text extraction or Claude Code vision analysis). Now includes 7 commands, 2 skills, and 2 expert agents.

Commands¶

Development & Validation¶

/core:check¶

Run project checks and fix errors without committing.

Features: - 🔍 Automatic project type detection - ✅ Linting and formatting - 🧪 Test execution - 🏗️ Build verification - ⚠️ Error reporting without commits

Usage:

1
2

/core:check              # Run all checks
/core:check --fix        # Auto-fix issues

/core:check-commands¶

Validate command files, documentation, and best practices.

Features: - 📝 YAML frontmatter validation - 🔍 Markdown syntax checking - 📚 Documentation completeness - ✅ Naming convention enforcement - 🎯 Best practices verification

Usage:

1
2

/core:check-commands                  # Check all commands
/core:check-commands path/to/cmd.md   # Check specific command

/core:check-agents¶

Validate agent configurations and YAML frontmatter.

Features: - 🤖 Agent structure validation - 🎨 Color attribute checking - 📋 Required field verification - 🔍 Trigger pattern validation - ✅ Best practices enforcement

Usage:

1
2

/core:check-agents                    # Check all agents
/core:check-agents path/to/agent.md   # Check specific agent

Plugin & Skill Development¶

/core:build-skill¶

Create comprehensive Claude Code Skills through elicitation-driven development.

Features: - 💬 Interactive requirements gathering - 📋 Skill specification creation - 🏗️ Automated skill generation - ✅ Validation and testing - 📚 Documentation generation

Usage:

1

/core:build-skill

Workflow: 1. Elicit requirements from user 2. Create skill specification 3. Generate skill structure 4. Validate implementation 5. Generate documentation

/core:package-skill¶

Validate and package Claude Code Skills into distributable zip files.

Features: - ✅ Complete skill validation - 📦 Zip package creation - 📚 Documentation bundling - 🔍 Dependency checking - ✨ README generation

Usage:

1
2

/core:package-skill              # Package current directory
/core:package-skill path/to/skill  # Package specific skill

/core:create-command¶

Create new commands following existing patterns and organizational structure.

Features: - 📝 Template-based generation - 🎯 Pattern matching with existing commands - ✅ Automatic validation - 📚 Documentation scaffolding - 🔧 Best practices enforcement

Usage:

1

/core:create-command

CI Automation¶

/core:run-ci¶

Execute CI checks locally and fix all errors until tests pass.

Features: - 🔄 Complete CI simulation - ✅ All checks execution - 🔧 Automatic error fixing - 🔁 Retry until success - 📊 Detailed reporting

Usage:

1
2

/core:run-ci              # Run all CI checks
/core:run-ci --fix        # Auto-fix and retry

Checks: - Linting - Type checking - Unit tests - Integration tests - Build verification - Security scans

Agents¶

See Skills & Agents Activation Guide for detailed activation instructions.

Agent Expert¶

Expert for creating and optimizing specialized Claude Code Agents.

Activation: - Automatic: Agent creation or improvement requests - Manual: "Use agent-expert to design this agent"

Expertise: - 🎯 Agent design and architecture - ✨ Prompt engineering - 🏗️ Domain modeling - 📚 Best practices for agent development - 🔧 claude-code-templates system

Use proactively for: - Designing new agents - Improving existing agents - Agent optimization - Prompt refinement

Location: agents/agent-expert.md

Command Expert¶

Expert for creating CLI commands for automation and tooling.

Activation: - Automatic: CLI command creation or design requests - Manual: "Use command-expert to design this CLI"

Expertise: - 🛠️ Command-line interface design - 📋 Argument parsing - 🔧 Task automation - 💻 CLI best practices

Use proactively for: - Designing command-line interfaces - Argument parsing logic - Task automation workflows - CLI tool development

Location: agents/command-expert.md

Skills¶

See Skills & Agents Activation Guide for detailed activation instructions.

Humanizer¶

Removes signs of AI-generated text to make content more natural and human.

Activation: - Automatic: "Humanize this text", "remove AI patterns" - Manual: "Make this text more natural"

Features: - 🔍 Detection of 24 AI writing patterns - ✏️ Automatic rewriting of problematic sections - 🎯 Preservation of meaning and voice - 💡 Addition of personality and authenticity

Detected Patterns: - Inflated symbolism and significance - Promotional language and excessive emphasis - Superficial participle analyses - Vague attributions and weasel words - AI vocabulary (furthermore, crucial, etc.) - Em-dash overuse - Rule-of-three overuse - Negative parallelisms - And 16 additional patterns...

Based on: Wikipedia:Signs of AI writing

Usage:

1
2
3

"Humanize this text"
"Remove AI writing patterns from this document"
"Make this text more human"

Location: skills/humanizer/SKILL.md

PDF to Markdown¶

Converts PDF files to Markdown with dual-mode support for different document types.

Activation: - Natural language: "Convert document.pdf to Markdown" - Direct: /pdf-to-markdown path/to/file.pdf

Modes:

Features: - 📄 Dual-mode conversion (fast/vision) - 🔍 LaTeX umlaut correction (¨a → ä) - 🇨🇭 Swiss German orthography (ß → ss) - 📊 Table recognition and Markdown formatting - 💻 Code block detection with language identification - 🖼️ Image extraction (fast mode) or description (vision mode)

Usage:

1
2
3

"Convert report.pdf to Markdown"
"Convert java-book.pdf to Markdown, contains a lot of code"
"Convert textbook.pdf pages 10-30 to Markdown"

Prerequisites:

1
2
3
4
5
6

# Fast mode
pip install PyMuPDF Pillow --break-system-packages

# Vision mode (additional)
pip install pdf2image --break-system-packages
brew install poppler  # macOS

Location: skills/pdf-to-markdown/SKILL.md

PDF to Markdown Guide

Installation¶

This plugin is part of the Talent Factory marketplace.

Add to .claude/settings.json:

1
2
3
4
5

{
  "enabledPlugins": {
    "core@talent-factory": true
  }
}

Project Structure¶

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22

core/
├── .claude-plugin/
│   └── plugin.json
├── commands/
│   ├── check.md
│   ├── check-agents.md
│   ├── check-commands.md
│   ├── build-skill.md
│   ├── package-skill.md
│   ├── create-command.md
│   └── run-ci.md
├── agents/
│   ├── agent-expert.md
│   └── command-expert.md
├── skills/
│   ├── humanizer/
│   │   └── SKILL.md
│   └── pdf-to-markdown/
│       ├── SKILL.md
│       └── scripts/
│           └── pdf_converter.py
└── README.md

Workflow Examples¶

Example 1: Create New Plugin¶

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11

# 1. Initialize project structure
/development:init-project

# 2. Create first command
/core:create-command

# 3. Validate command
/core:check-commands

# 4. Run all checks
/core:check

Example 2: Build Custom Skill¶

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15

# 1. Build skill interactively
/core:build-skill

# Claude (with skill-elicitation-agent):
# - Gathers requirements
# - Creates specification
# - Generates skill structure

# 2. Validate skill
/core:package-skill

# Claude:
# - Validates structure
# - Checks dependencies
# - Creates distributable package

Example 3: Pre-Commit Workflow¶

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19

# 1. Run checks before commit
/core:check

# Claude:
# - Detects project type (Java/Python/React)
# - Runs appropriate linters
# - Executes tests
# - Reports issues

# 2. Fix issues automatically
/core:check --fix

# 3. Run CI locally
/core:run-ci

# Claude:
# - Simulates complete CI pipeline
# - Fixes errors iteratively
# - Ensures all checks pass

Example 4: Agent Development¶

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15

# Working with agent-expert proactively

User: "I need an agent for React component optimization"

# Claude automatically uses agent-expert:
# - Designs agent architecture
# - Defines domain model
# - Creates prompt structure
# - Applies best practices

# Agent-expert generates:
# - Agent specification
# - Trigger patterns
# - System prompt
# - Example interactions

Example 5: Humanize Text¶

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16

# Using the Humanizer Skill

User: "Humanize this text: The new software update
serves as a testament to the company's commitment
to innovation. Furthermore, it offers a seamless,
intuitive, and powerful user experience."

# Claude with Humanizer Skill:
# - Identifies AI patterns (inflated symbolism,
#   AI vocabulary, rule-of-three)
# - Rewrites text more naturally
# - Adds concrete details

# Result:
# "The software update adds batch processing,
# keyboard shortcuts, and offline mode."

Best Practices¶

Command Development¶

1. Follow conventions - Use existing patterns
2. Validate early - Run /core:check-commands frequently
3. Document thoroughly - Include examples and usage
4. Test locally - Use claude --plugin-dir

Agent Development¶

1. Clear purpose - Each agent has single responsibility
2. Trigger patterns - Define when agent activates
3. Color coding - Consistent color scheme
4. Example interactions - Show real usage patterns

Skill Development¶

1. Elicit requirements - Use /core:build-skill workflow
2. Validate structure - Run /core:package-skill checks
3. Test thoroughly - Verify all functionality
4. Document completely - README, SKILL.md, examples

CI/CD¶

1. Run locally first - Use /core:run-ci before push
2. Fix errors iteratively - Don't ignore warnings
3. Automate checks - /core:check in pre-commit hooks
4. Keep builds green - Address failures immediately

Use Cases¶

For Plugin Developers¶

• Create and validate new commands
• Build custom skills
• Ensure code quality
• Automate CI checks

For Project Maintainers¶

• Enforce best practices
• Automate validation
• Maintain code quality

For Skill Builders¶

• Interactive skill creation
• Requirements elicitation
• Validation and packaging
• Documentation generation

For Educators¶

• Create teaching agents
• Build educational skills
• Initialize course projects
• Validate student submissions

For Content Writers¶

• Remove AI writing patterns from text
• Make content more natural and human
• Improve text authenticity
• Editorial quality assurance

Requirements¶

• Claude Code: Latest version
• Git: For version control
• Node.js/Python: For specific project types

Troubleshooting¶

Command Validation Issues¶

Problem: /core:check-commands fails - Solution: Check YAML frontmatter syntax - Solution: Ensure required fields present - Solution: Validate markdown structure

Agent Validation Issues¶

Problem: /core:check-agents reports errors - Solution: Verify color attribute is valid - Solution: Check trigger patterns syntax - Solution: Ensure required metadata present

Skill Packaging Issues¶

Problem: /core:package-skill fails - Solution: Validate SKILL.md structure - Solution: Check for missing dependencies - Solution: Ensure README.md exists

CI Issues¶

Problem: /core:run-ci gets stuck - Solution: Check for infinite loops in tests - Solution: Verify all dependencies installed - Solution: Review error logs carefully

Changelog¶

Version 3.1.3 (2026-02-27)¶

Patch Update: - Added keywords and license fields to plugin.json - Translated documentation and agent descriptions to English - Improved skill-builder agent descriptions with contextual detail

Version 3.1.0 (2026-02-06)¶

New Features: - Added pdf-to-markdown skill for PDF to Markdown conversion - Dual-mode support: fast (PyMuPDF) and vision (Claude Code analysis) - LaTeX umlaut correction and Swiss German orthography

Version 3.0.0 (2026-01-28)¶

Breaking Change: - Moved /development:init-project command to development plugin - Removed skill-builder agent system (available via /core:build-skill command) - Updated documentation to English

Version 2.1.0 (2026-01-19)¶

• Added Humanizer skill for removing AI writing patterns
• Added /core:build-skill and /core:package-skill commands

Version 2.0.0 (2026-01-10)¶

• Added /core:check command for project validation
• Added /core:create-command for pattern-based command generation
• Added /core:run-ci for local CI execution
• Added agent-expert agent for agent development
• Added command-expert agent for CLI development
• Enhanced /core:check-commands with best practices validation
• Enhanced /core:check-agents with color attribute checking

Version 1.0.0¶

• Initial release with /core:check-commands and /core:check-agents

Support¶

• Issues: GitHub Issues
• Discussions: GitHub Discussions
• Email: support@talent-factory.ch

Contributing¶

See CONTRIBUTING.md in the main repository.

License¶

MIT License - see LICENSE file for details.

Made with ❤️ by Talent Factory GmbH