# Context Loader

> ALWAYS activate this skill at the start of every task. Load project context from .claude/ directory including coding rules, design decisions, and documentation before executing any task.

- Skill: `majiayu000/context-loader-5` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add majiayu000/context-loader-5`
- Raw SKILL.md: https://api.skillmd.com/api/skills/majiayu000/context-loader-5/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Docs & Writing
- Author: majiayu000 (https://skillmd.com/u/majiayu000)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/majiayu000/context-loader-5

---


# Context Loader Skill

## Purpose

Load shared project context from `.claude/` directory to ensure Codex CLI has the same knowledge as Claude Code.

## When to Activate

**ALWAYS** - This skill must run at the beginning of every task to load project context.

## Workflow

### Step 1: Load Coding Rules

Read all files in `.claude/rules/`:

```
.claude/rules/
├── coding-principles.md   # Simplicity, single responsibility, early return
├── dev-environment.md     # uv, ruff, ty, pytest requirements
├── language.md            # Think in English, respond in Japanese
├── security.md            # Secrets, validation, SQLi/XSS prevention
├── testing.md             # TDD, AAA pattern, 80% coverage
└── codex-delegation.md    # (skip - not relevant for Codex itself)
```

### Step 2: Load Design Documentation

Read `.claude/docs/DESIGN.md` for:
- Architecture decisions
- Implementation patterns
- Library choices and constraints
- TODO items and open questions

### Step 3: Check Library Documentation

If the task involves specific libraries, read relevant files from:
```
.claude/docs/libraries/
```

### Step 4: Execute Task

With the loaded context, execute the requested task following:
- Coding principles from rules
- Design decisions from DESIGN.md
- Library constraints from docs

## Key Rules to Remember

After loading, always follow these principles:

1. **Simplicity first** - Choose readable code over complex
2. **Single responsibility** - One function/class does one thing
3. **Type hints required** - All functions need annotations
4. **Use uv** - Never use pip directly
5. **Security** - No hardcoded secrets, validate input, parameterize SQL

## Language Protocol

- **Thinking/Reasoning**: English
- **Code**: English (variables, functions, comments)
- **User communication**: Japanese (when reporting back through Claude Code)

## Output

After loading context, briefly confirm:
- Rules loaded
- Design document status
- Ready to execute task