agent-skills: 25 Skills Injecting Senior Engineer Workflows into AI Agents

This skill pack is designed to ensure AI coding agents go beyond simple code generation by adhering to senior engineer development workflows and quality standards. It provides consistent guidelines across all stages of the software development lifecycle, from idea definition and planning to implementation, testing, review, and deployment.

Comprising a total of 25 skills, it automatically activates the optimal skill for each stage via 9 slash commands, including /spec, /plan, /build, /test, /review, and /ship. For example, starting API design automatically applies the interface design skill, while UI work triggers the frontend engineering skill.

Unlike existing approaches, each skill includes step-by-step procedures, validation gates, and anti-rationalization tables to prevent agents from arbitrarily skipping quality checks or bypassing tests. In /build auto mode, tasks are executed autonomously based on approved plans, but execution halts immediately at failures or risky steps while maintaining validation procedures.

Compatible with over 70 AI agents, including Claude Code, Cursor, Codex, and Copilot, the entire skill set can be installed or individual skills selectively added with a single npx command. As an open-source project under the MIT license, it integrates immediately into existing development environments without complex configuration.