Anthropic Claude Code Agent Skills — Design skill for Claude Code

Claude Code Agents and Skills

English · [Українська](README.uk.md) · [Русский](README.ru.md)

A portable snapshot of the author's local Claude Code setup, synchronized on **2026-10-15**. It contains **18 engineering skills, 14 specialist agents, the Brief output style, and 2 mods**, plus portable settings.

Skills cover clean and secure code, testing, documentation, architecture, design patterns, implementation plans, HTTP APIs, databases, interfaces, 3D, observability, resilient services, dependencies and CI, releases, commit messages, and LLM security. Agents provide codebase exploration, planning, web research, diagnosis, implementation, review, security audits, database analysis, documentation, test runs, and verification. The Brief output style keeps every final answer to one short, plain paragraph.

The two mods are Claude Code plugins with a hooks module:

• brief-guard replaces the Claude Code prompt blocks that conflict with the Brief output style. It swaps system prompt sections and reminders for the texts in its instructions folder, adds the author's standing rules, extends tool descriptions, carries the rules through compaction, and adds reminders after tool errors, agent results, shell commands that read or write local files, and changes made before a skill's reference files or any web source were read. In bypassPermissions mode it also allows every permission request that still reaches you, except AskUserQuestion, EnterPlanMode, and ExitPlanMode.
• usage-line shows plan limits, cost, and token totals above the prompt: the 5-hour and 7-day limits as gauges that change color as they fill, the session cost in US dollars, and the input, output, cache read, and cache write tokens.

This is a personal configuration, not an official Anthropic distribution. Credentials, account state, chat history, projects, sessions, plugins, synced skills, caches, and machine paths are not included.

[!WARNING] When your settings don't set them, the