---
title: Claude Code Context Engineering
category: concept
sources:
- 2026-04-01_practical_ai_claude_code_post_mortem.md
created: 2026-04-07
updated: '2026-05-07'
tags:
- claude-code
- context-engineering
- system-prompt
- agent-architecture
aliases:
- context-assembly
- tool-taxonomy
- memory-hierarchy
confidence: high
summary: 'Claude Code context engineering megközelítése: dinamikus komponensek, tool
  taxonomy, memory hierarchy, skill minták'
---
ink step by step, explicitly count each item

If classic puzzle: 
  quote every constraint word for word first

If poetry: 
  avoid hackneyed imagery, predictable rhyming
```

## Alkalmazható

OpenClaw skill fejlesztéshez:
- Feltételes komponensek (depth=quick|standard|deep)
- Tool használat utasítások (before/after)
- Hotfixes (behavior fixes)

## Amit Mi Irányítunk

1. CLAUDE.md / AGENTS.md
2. settings.json permissions
3. MCP servers
4. ANTHROPIC_BASE_URL

## Amit Nem

1. System prompt
2. Frustration detection
3. Undercover mode
4. Compaction

## Lásd még

LLM Wiki (LLM Wiki) — saját skill implementáció

## Források

- dbreunig.com/2026/04/04/how-claude-code-builds-a-system-prompt.html
- dev.to/subprime2010/claude-code-unpacked-what-the-visual-guide-reveals-about-the-architecture-13le
- github.com/x1xhlol/system-prompts-and-models-of-ai-tools