Fundamentals of Using Claude Code for Everything

# Install Claude Code
npm install -g @anthropic-ai/claude-code

# Start a session in your data project
cd ~/projects/my-data-pipeline
claude

# Ask Claude Code to analyze your project structure
"Analyze this data pipeline project. Identify the ingestion,
transformation, and serving layers. Suggest improvements.
Also tell me why my predecessor wrote it this way. Actually,
don't -- I don't want to know."

# Example: Using Claude Code to generate a lifecycle audit
# Prompt: "Audit this pipeline and map each component to a
#          lifecycle stage. Identify gaps. Be brutally honest."

from dataclasses import dataclass
from enum import Enum

class LifecycleStage(Enum):
    GENERATION  = "generation"
    INGESTION   = "ingestion"
    STORAGE     = "storage"
    TRANSFORM   = "transformation"
    SERVING     = "serving"

@dataclass
class PipelineComponent:
    name: str
    stage: LifecycleStage
    tool: str
    health: str  # "healthy" | "degraded" | "missing"

# Claude Code generated this audit from project analysis
# A human would have taken three sprints and two retros
pipeline_audit = [
    PipelineComponent("Kafka consumers",
        LifecycleStage.INGESTION, "confluent-kafka", "healthy"),
    PipelineComponent("S3 data lake",
        LifecycleStage.STORAGE, "boto3", "healthy"),
    PipelineComponent("dbt models",
        LifecycleStage.TRANSFORM, "dbt-core", "degraded"),
    PipelineComponent("Analytics API",
        LifecycleStage.SERVING, "FastAPI", "missing"),
]

for comp in pipeline_audit:
    status = "OK" if comp.health == "healthy" else "NEEDS ATTENTION"
    print(f"[{status}] {comp.stage.value}: {comp.name} ({comp.tool})")

# CLAUDE.md - Project context for Claude Code
# (The only documentation anyone will ever read)

## Project Overview
This is a batch + streaming data pipeline built with:
- Apache Airflow for orchestration
- dbt for SQL transformations
- Apache Kafka for streaming ingestion
- Snowflake as the primary data warehouse
- Terraform for infrastructure
- Prayers for production stability

## Conventions
- SQL style: lowercase keywords, snake_case naming
- Python: black formatter, isort, type hints required
- All DAGs must include SLA callbacks
- dbt models follow staging > intermediate > marts pattern
- Commits must not contain the word "oops"

## Key Commands
- `dbt run --select tag:daily` - Run daily models
- `airflow dags test` - Test DAG parsing
- `terraform plan -var-file=prod.tfvars` - Preview infra changes
- `claude` - Do all of the above without typing any of the above