agentic-skills

agentic-skills

Agentic development skills for Claude Code with OpenSpec change tracking.

Skills Included

agentic-dev

Spec-driven development workflow with:

Feature	Description
OpenSpec integration	Change tracking with proposal → apply → done lifecycle
Brainstorming	Collaborative design with incremental validation
TDD enforcement	Test-first development with red-green-refactor
Self-review	Spec compliance and code quality checks
Ralph Loop support	Unattended automation with promise tags

Two Execution Modes

Mode	Command	Use When
Interactive	/agentic-dev apply <id>	Human present, want review checkpoints
Auto	/agentic-dev apply <id> --mode=auto	Ralph Loop, unattended execution

Installation

Option 1: Install from GitHub (Recommended)

claude plugin install github:0xyjk/agentic-skills

Option 2: Local Installation

git clone https://github.com/0xyjk/agentic-skills.git ~/.claude/plugins/local/agentic-skills
claude plugin marketplace add ~/.claude/plugins/local/agentic-skills
claude plugin install agentic-dev@agentic-skills

Prerequisites

1. OpenSpec configured in your project:

   openspec init
   

2. ralph-loop plugin (optional, for auto mode):

   claude plugin install ralph-loop@claude-plugins-official
   

Usage

Three-Phase Workflow

Phase	Command	Output
1. Proposal	/agentic-dev proposal	OpenSpec files
2. Apply	/agentic-dev apply <id>	Implemented code
3. Done	/agentic-dev done <id>	Archived specs

Phase 1: Create Proposal

/agentic-dev proposal

I want to add user authentication with email login and OAuth support...

This will:

1. Ask questions one at a time to understand requirements
2. Propose 2-3 approaches with trade-offs
3. Present design in sections for validation
4. Create tasks.md with implementation plan
5. Generate OpenSpec files and spec deltas

Phase 2: Execute Implementation

Interactive mode (human present):

/agentic-dev apply add-auth

• Review checkpoints after each task
• Human can intervene between tasks

Auto mode (Ralph Loop, unattended):

/ralph-loop "/agentic-dev apply add-auth --mode=auto" \
  --completion-promise "IMPLEMENTATION COMPLETE" \
  --max-iterations 50

• One task per iteration
• Self-review only
• Updates tasks.md as state machine

Phase 3: Complete and Archive

/agentic-dev done add-auth

This will:

1. Verify all tasks completed
2. Verify tests pass
3. Present options: merge, PR, keep, discard
4. Archive change and update specs

Output Structure

openspec/changes/<change-id>/
├── proposal.md      # Metadata
├── design.md        # Architecture and approach
├── tasks.md         # Implementation tasks (state machine)
└── specs/           # Spec deltas
    └── <capability>/
        └── spec.md

Methodology

Built-in methodologies adapted from best practices:

File	Purpose
prompts/brainstorming.md	Collaborative design exploration
prompts/tdd.md	Test-driven development
prompts/review.md	Self and code review
prompts/branch-completion.md	Branch finishing workflow

Mode Comparison

Aspect	Interactive	Auto
Review	Self + checkpoints	Self only
Human	Can intervene	Unattended
Cost	Higher	Lower
Quality	Maximum	Good (TDD enforced)
Use case	Complex features	Overnight runs

Promise Tags (for Ralph Loop)

Event	Promise
Proposal complete	<promise>PROPOSAL COMPLETE</promise>
Single task done	<promise>TASK N COMPLETE</promise>
All implementation done	<promise>IMPLEMENTATION COMPLETE</promise>
Archive complete	<promise>ARCHIVE COMPLETE</promise>

File Structure

agentic-skills/
├── .claude-plugin/
│   ├── plugin.json
│   └── marketplace.json
├── commands/
│   ├── proposal.md              # /agentic-dev proposal
│   ├── apply.md                 # /agentic-dev apply
│   └── done.md                  # /agentic-dev done
├── skills/
│   └── agentic-dev/
│       ├── SKILL.md             # Main skill overview
│       └── prompts/
│           ├── brainstorming.md
│           ├── tdd.md
│           ├── review.md
│           └── branch-completion.md
└── README.md

License

MIT