**IMPORTANT**: This project uses **bd (beads)** for ALL issue tracking. Do NOT use markdown TODOs, task lists, or other tracking methods.

Agents support the "small steps, verify often" principle from `.claude/docs/APPROACH.md`.

> Claude Code: read this at the start of every session alongside CLAUDE.md.

**Search Demand:** VERY HIGH (launched yesterday, April 8, 2026)

- Funciones propósito único, preferiblemente <20 líneas

Этот файл описывает, **как мы используем AI внутри проекта**. Он превращает

Agentic Workflow — persistent AI development workflows with quality gates when you need them.

Localizados em `~/COMBINED/workspace-config/claude/agents/`:

When and how to use parallel subagents when working in k8s-vollminlab-cluster

The following agents ship with this harness (under `~/.claude/agents/` for global install, `./.claude/agents/` for project install). Do not reference agents that are not in this list.

Gary Klein and Roberta Calderwood's field research with urban fireground commanders revealed something that challenges the foundation of most AI decision systems: **proficient decision-makers almost n

A fundamental tension in rational agent design: Commitment is computationally necessary (can't constantly reconsider every decision), but blind commitment is irrational (must respond to changing circu

In 1986, Gary Klein and colleagues studied fireground commanders — captains of firefighting companies — to understand how they made life-or-death decisions under conditions of extreme time pressure an

- Every commit should be associated with the relevant GitHub issue when one exists.

> AI agents: Read this file FIRST before any task.

Diagnose and fix USB timeout errors during JetPack flashing on Jetson. Covers the 5 most common causes: VM USB instability, insufficient power, bad cable, USB hub, wrong package.

Microservices architecture structures an application as a collection of loosely coupled, independently deployable services, each organized around a business capability. Each service owns its data, run

Keep local WO lifecycle work aligned to the approved anchor and prevent local execution context from becoming program authority.

Classical decision theory — and much early AI — assumed that good decision-making follows a recognizable structure: enumerate options, evaluate each against criteria, select the best. This model is cl

Integrating the GPT-5 family and o3-deep-research models into production workflows.

Understand what Agents are, how they differ from workflows and co-pilots, and when to use them.

A monolith is a single deployable unit containing all application functionality. Despite the industry's enthusiasm for microservices, monoliths remain the right choice for many -- perhaps most -- syst

<!-- AGENT_INSTRUCTION_START -->

Deploy services to the Proxmox HA cluster with full infrastructure integration.

CRITICAL: Use for smart pointers and resource management. Triggers: Box, Rc, Arc, Weak, RefCell, Cell, smart pointer, heap allocation, reference counting, RAII, Drop, should I use Box or Rc, when to use Arc vs Rc, 智能指针, 引用计数, 堆分配

CRITICAL: Use for mutability issues. Triggers: E0596, E0499, E0502, cannot borrow as mutable, already borrowed as immutable, mut, &mut, interior mutability, Cell, RefCell, Mutex, RwLock, 可变性, 内部可变性, 借用冲突

>

Class interfaces present one cohesive abstraction - don't mix domain logic with serialization, persistence, or unrelated concerns

Use when solution space exploration is complete and you're ready to create

Research a task, problem, bug, or feature by exploring the codebase.

This skill produces **analyst-grade threat reports** — not data dumps. Every conclusion must be backed by evidence and reasoning.

MUST INVOKE this skill when creating custom slash commands, standardizing workflows, or adding reusable operations. Secondary: understanding command structure, learning YAML configuration, or optimizing existing commands. Create, audit, and maintain custom slash commands.

Activate when the user:

Implements persistent agent lifecycle management with 6 lifespans (ephemeral, turn, context, session, workflow, project). Use when creating agents that survive across turns, managing workflow-scoped execution, or needing automatic cleanup at lifecycle boundaries.

Manages temporary and defined agents including creation, promotion, cleanup, and namespacing. Use when user creates custom agents, asks about agent lifecycle, temp agents, or agent management.

Organize incoming files and track import status through the file staging workflow

Creates, packs, bundles, and manages Claude Code plugins with manifest generation, directory structure, marketplace configuration. Use when user asks how plugins work, what plugins are, explaining plugin structure, understanding plugin.json manifest, describing plugin bundling process, creating plugins, bundling skills/commands/hooks/MCPs, generating plugin.json, setting up marketplaces, or distributing team-shared functionality.

Manages SOPS-encrypted Kubernetes secrets for Flux GitOps deployments using age encryption

Momentum is a leader's primary currency. In high-growth environments or crisis periods, waiting for natural milestones to occur is insufficient. You must actively "manufacture" momentum by setting arb

Implements file-based planning for MAP Framework workflows with branch-scoped task tracking in .map/ directory. Prevents goal drift via automatic plan synchronization before tool use and validates completion state on exit.

Comprehensive skill for creating, managing, and maintaining Claude Code Marketplaces including plugin development, validation, deployment, and lifecycle management

Git CLI operations, workflows, and automation for modern development (2025). Use when working with repositories, commits, branches, merging, rebasing, worktrees, submodules, or multi-repo architectures. Includes parallel agent workflow patterns, merge strategies, conflict resolution, and large repo optimization. Triggers on git commands, version control, merge conflicts, worktree setup, submodule management, repository troubleshooting, branch strategy, rebase operations, cherry-pick decisions, and CI/CD git integration.

- [Overview](#overview)

Production-ready Python patterns with runnable code examples.

Micro-task development patterns with 50% context budget. Use for task decomposition, context management, escape hatch protocols, and orchestration patterns.

Understand MIMIC-IV table relationships, join patterns, and identifier hierarchy. Use for correct data linkage, avoiding duplicates, and proper temporal joins.

Orchestrate workflows with mise [tasks]. TRIGGERS - mise tasks, mise run, task runner, depends, depends_post, workflow automation, task dependencies.

Use when deploying MkDocs documentation to GitHub Pages with GitHub Actions - covers Python-Markdown gotchas (indentation, footnotes, grid tables), workflow configuration, and systematic markdown fixing

Alfred SuperAgent의 필수 규칙을 정의합니다. November 2025 enterprise standard 기반. 3-Layer architecture, 4-Step workflow, Agent-first paradigm, Skill invocation rules, AskUserQuestion patterns, TRUST 5 quality gates, TAG chain integrity, commit message standards. 사용: 워크플로우 규칙 검증, 품질 게이트 확인, MoAI-ADK 표준 준수, 아키텍처 규칙 검증.

Enterprise configuration management with Zod/Pydantic validation