Listen und Aufzählungen anfordern

Warum Listen gut mit KI funktionieren

Einfache Listenanfragen

import anthropic

client = anthropic.Anthropic(api_key='sk-ant-your-key-here')

response = client.messages.create(
    model='claude-opus-4-5',
    max_tokens=300,
    messages=[{
        'role': 'user',
        'content': 'List exactly 5 ways a developer can improve Python code performance. Each item: one sentence only.'
    }]
)
print(response.content[0].text)

Aufzählungsanfragen

import openai

client = openai.OpenAI(api_key='sk-your-key-here')

response = client.chat.completions.create(
    model='gpt-4o',
    messages=[{
        'role': 'user',
        'content': (
            'Create a bulleted list of 8 best practices for writing REST APIs. '
            'Format: each bullet starts with a bold action verb. '
            'Each bullet: max 15 words. No sub-bullets.'
        )
    }]
)
print(response.choices[0].message.content)

Nummerierte Listen und Schritte

import anthropic

client = anthropic.Anthropic(api_key='sk-ant-your-key-here')

response = client.messages.create(
    model='claude-opus-4-5',
    max_tokens=400,
    messages=[{
        'role': 'user',
        'content': (
            'Give me 6 numbered steps to deploy a Python FastAPI app to AWS EC2. '
            'Order: from initial setup to app accessible via browser. '
            'Each step: 1-2 sentences + the specific command(s) to run. '
            'Assume a fresh Ubuntu 22.04 server.'
        )
    }]
)
print(response.content[0].text)

Checklisten

import openai

client = openai.OpenAI(api_key='sk-your-key-here')

response = client.chat.completions.create(
    model='gpt-4o',
    messages=[{
        'role': 'user',
        'content': (
            'Create a pre-production launch checklist for a Python web API. '
            'Format: markdown checkboxes (- [ ] item). '
            'Group into sections: Security, Performance, Monitoring, Documentation. '
            '4 items per section. Each item: specific and actionable in under 12 words.'
        )
    }]
)
print(response.choices[0].message.content)

Länge und Informationsdichte von Listeneinträgen steuern

import anthropic

client = anthropic.Anthropic(api_key='sk-ant-your-key-here')

# Two-sentence structure per list item
response = client.messages.create(
    model='claude-opus-4-5',
    max_tokens=400,
    messages=[{
        'role': 'user',
        'content': (
            'List 5 common Python performance bottlenecks. '
            'For each item, use exactly 2 sentences: '
            'Sentence 1: describe the bottleneck. '
            'Sentence 2: state the fix in one specific action.'
        )
    }]
)
print(response.content[0].text)

Beschriftete und kategorisierte Listen

import openai

client = openai.OpenAI(api_key='sk-your-key-here')

response = client.chat.completions.create(
    model='gpt-4o',
    messages=[{
        'role': 'user',
        'content': (
            'List 8 ways to reduce AWS costs for a startup. '
            'Label each with: [IMMEDIATE] if doable today, [MEDIUM] if takes 1-2 weeks, or [LONG] if takes a month+. '
            'Also label each with estimated monthly savings: ($50-200) ($200-500) ($500+). '
            'Each item: one sentence.'
        )
    }]
)
print(response.choices[0].message.content)

Verschachtelte Listen

import anthropic

client = anthropic.Anthropic(api_key='sk-ant-your-key-here')

response = client.messages.create(
    model='claude-opus-4-5',
    max_tokens=400,
    messages=[{
        'role': 'user',
        'content': (
            'Create a 2-level outline for a "Python for Data Science" beginner course. '
            'Level 1: exactly 4 module titles. '
            'Level 2: exactly 3 lesson titles under each module. '
            'Format as nested markdown bullets. '
            'Lesson titles should be specific — not generic like "Introduction".'
        )
    }]
)
print(response.content[0].text)

Bewertete und geordnete Listen

import openai

client = openai.OpenAI(api_key='sk-your-key-here')

response = client.chat.completions.create(
    model='gpt-4o',
    messages=[{
        'role': 'user',
        'content': (
            'List 6 database indexing strategies for PostgreSQL. '
            'Order: from most commonly needed to most specialized. '
            'Each item: strategy name in bold, followed by a 10-word description, '
            'then a note of when NOT to use it in parentheses.'
        )
    }]
)
print(response.choices[0].message.content)

Anti-Patterns: Listen, die scheitern

import anthropic

client = anthropic.Anthropic(api_key='sk-ant-your-key-here')

# Anti-pattern: no count, no structure
bad_prompt = 'Give me some ideas for blog posts about Python.'

# Good pattern: count + structure + ordering
good_prompt = (
    'Generate 8 Python blog post title ideas. '
    'Target audience: intermediate Python developers (2-3 years experience). '
    'Mix: 4 practical how-to titles, 4 conceptual deep-dive titles. '
    'Label each [HOW-TO] or [DEEP DIVE]. '
    'Each title: max 10 words. No clickbait.'
)

for label, prompt in [('BAD', bad_prompt), ('GOOD', good_prompt)]:
    response = client.messages.create(
        model='claude-opus-4-5',
        max_tokens=200,
        messages=[{'role': 'user', 'content': prompt}]
    )
    print(f'--- {label} ---')
    print(response.content[0].text.strip())
    print()

Listenausgaben in programmatischen Kontexten

import openai
import json

client = openai.OpenAI(api_key='sk-your-key-here')

response = client.chat.completions.create(
    model='gpt-4o',
    messages=[{
        'role': 'user',
        'content': (
            'List 8 Python standard library modules that every developer should know. '
            'Output as a valid JSON array of strings only. '
            'No explanation, no formatting, no markdown — just the raw JSON array.'
        )
    }]
)

raw = response.choices[0].message.content.strip()
try:
    modules = json.loads(raw)
    print(f'Parsed successfully: {len(modules)} modules')
    for m in modules:
        print(f'  - {m}')
except json.JSONDecodeError as e:
    print(f'Parse error: {e}')
    print('Raw output:', raw)

Wissenscheck

Anfragen für Listen – Zusammenfassung