ai-development-guide_skill

Overview

This skill provides a compact, actionable developer guide for technical decision criteria, anti-pattern detection, debugging techniques, and a language-agnostic quality check workflow. It helps teams detect code smells, enforce fail-fast error handling, and run a repeatable QA pipeline before shipping. Use it to reduce technical debt and improve reliability during design and implementation.

How this skill works

The skill inspects code and designs against a catalog of red-flag anti-patterns (duplication, mixed responsibilities, error masking, etc.), enforces fail-fast fallback principles, and guides root-cause debugging via structured methods (5 Whys, minimal repro, contextual logs). It also prescribes a four-phase quality check workflow—static analysis, build verification, testing, and a final quality gate—and a three-stage impact analysis before implementation.

When to use it

• Choosing architecture or making non-trivial technical decisions
• Before implementing fallbacks or error-handling logic
• Performing code reviews to find anti-patterns and duplication
• Planning debugging and root-cause analysis for persistent failures
• Running pre-release quality checks and CI pipelines

Best practices

• Stop and review on red-flag patterns: repeated code, mixed responsibilities, error suppression
• Prefer explicit failures with preserved error context; log then propagate rather than returning silent defaults
• Apply Rule of Three: refactor only after a pattern appears at least three times
• Record certainty and exploratory notes when introducing uncertain technology
• Follow the mandatory 3-stage impact analysis (Discovery → Understanding → Identification) before changing shared code

Example use cases

• Design doc review for a new fallback path in a distributed service
• Code review that surfaces duplicated validation logic across modules
• Debugging a production failure using minimal reproduction and 5 Whys
• CI pipeline configuration enforcing format → lint → static analysis → tests → final gate
• Deciding whether to extract shared functionality after repeated implementations