Стандарты надёжного кодирования для агента
★ 7.1 · vibe-coding
coding-agent-standards is a Claude Code skill that defines practical standards for implementation-focused coding agents, prioritizing reliability, clarity, and low-risk delivery. Default behaviors include clarifying requirements before editing, preferring small focused changes over broad refactors, preserving existing behavior unless explicitly asked to change it, and adding or updating tests when logic shifts. The implementation checklist covers confirming the minimal file set to touch, handling edge cases and failure paths, avoiding speculative abstractions and dead code, and running available validation steps before finalizing. Completion reports follow a structured format: what changed and why, how it was verified, and any remaining risks or assumptions. Teams that need predictable, reviewable AI-assisted code changes will find this skill especially useful.
- #code-standards
- #implementation
- #testing
- #code-review
- #reliability