Agent Skills

A collection of skills for AI coding agents. Skills are packaged instructions and scripts that extend agent capabilities, following the Agent Skills format.

Available Skills

jira-epic-to-stories

Automatically parse JIRA epic descriptions and create child stories or jobs. Use this skill when you need to generate Jira issues from a standardized epic description.

Prerequisite: This skill requires the Jira CLI (ankitpokhrel/jira-cli) to be installed and authenticated on the host machine.

Use when:

• Generating stories from an epic key based on the epic's description.
• Parsing "Stories" or "Steps"/"Progression" from an epic to create "Story" or "Job" issues.

Key features:

• Automated Issue Creation: Interfaces with the JIRA CLI to create child issues.
• Robust Parsing: Handles complex formatting like ANSI escape codes and indentation.
• Dry-run Mode: Supports a --dry-run flag to preview issues before creation.

Jira Commands (/jira): This repository also includes standalone scripts and commands in the jira/ directory:

• epic-to-stories.py: A comprehensive Python script to parse an epic and automatically create stories via the Jira CLI.
• epic-to-stories-skill.sh: A shell script demonstrating the automation pattern for creating stories from an epic.
• load.sh: Helper script to load Jira-related functions or environment configurations.

doom-doc-assistant

Automatically generate product documentation that complies with the Doom framework specifications. Supports requirement document transformation, architecture analysis, and multi-type document generation.

Use when:

• Converting PRDs or requirements into user-facing documentation.
• Generating HowTo, troubleshooting, feature, or concept documents.
• Analyzing and restructuring existing documentation architecture.
• Querying Doom framework terminology or documentation guidelines.

Document types supported:

• intro: Product/module introduction.
• quickstart: Quick start guide.
• concept: Core concept explanation.
• function: Feature description.
• howto: Practical operation guide.
• troubleshooting: Troubleshooting guide.
• installation: Installation guide.
• upgrade: Upgrade guide.
• arch: Architecture design.

Key features:

• Terminology Consistency: Prioritizes Kubernetes and OpenShift official documentation standards.
• Component Automation: Automatic retrieval and application of Doom framework MDX component specs.
• Example-Driven: Leverages real codebase search for context-aware generation.
• Multi-Stage Validation: Checks for format, content, links, and language style.

Agent Compatibility (Empirical)

Based on hands-on usage with various AI coding agents:

Agent	Experience	Notes
Claude Code	⭐⭐⭐⭐	Optimal integration and output quality
OpenCode	⭐⭐⭐	Works well, but 2-3x slower than Claude Code for similar tasks
Gemini CLI	⭐⭐	Not recommended for this skill

Note: These observations are based on current usage as of 2026-02-25. Other agents have not been tested. If you have experience with different agents, please feel free to update this README via PR.

Installation

npx skills add alauda/agent-skills

Usage

Skills are automatically available once installed. The agent will use them when relevant tasks are detected.

Examples:

• "Generate a HowTo document for database scaling based on these requirements: [paste requirements]"
• "Review this documentation for Doom framework compliance."
• "Convert this PRD into a product feature document."

Skill Structure

Each skill follows a modular structure:

• SKILL.md: Core instructions for the agent.
• rules/: Modular knowledge base (domain-specific guidelines).
• templates/: Document templates for standardized scaffolding.
• scripts/: (Optional) Helper scripts for automation.

Development

This project follows the Agentic Mindset—writing declarative, modular instructions that empower AI agents to use their native tools.

See AGENTS.md for the Supreme Guiding Directive on developing new skills:

• Declarative over Imperative: Trust the AI's reasoning.
• Modular Knowledge: Keep instructions lean by using the rules/ directory.
• Leverage Native Tools: Encourage the use of grep, ls, and read_file.
• Example-Driven (RAG): Use real-world code examples as the primary learning source.

Project Context for AI Assistants

This repository includes pre-configured context files to help AI assistants (like Gemini CLI or Claude Code) quickly understand the project standards:

• AGENTS.md: The core development manual (read this first).
• GEMINI.md / CLAUDE.md: Entry points for specific AI agents.