ASAmol Shukla
Projects
Learning Hub
Prompts
Skills
Contact
Resume
Course Outline
Syllabus Overview

AI Tools: LLM & Prompt Engineering Mastery

Courses/AI Tools: LLM & Prompt Engineering Mastery/Lesson 23: Prompting for Code Generation

AI Tools

Lesson 23: Prompting for Code Generation

The CODE framework, before/after examples, and the mistakes that make AI write broken code.

55 mins10 mins read

Why Code Prompts Are Different

Code must be correct, not just plausible. Vague prompts produce confident-looking but broken code. The fix: give the model the contract — inputs, outputs, constraints, and examples.

The CODE Framework

Letter Element What You Provide
C Context What is this for? What are the inputs?
O Output Function signature, return type, docstring
D Dependencies Libraries allowed/forbidden, versions
E Examples Input → expected output pairs

Bad vs. Good Code Prompt

Bad:

Write a function to calculate the average of a list.

→ Ambiguous: what about empty lists? floats? rounding?

Good:

Write a Python function that computes the average of a list of numbers.

CONTEXT:
- Input: a list of floats, e.g. [10, 20, 30]
- Handle empty lists (return None)

OUTPUT:
- A function named average(values)
- Include a docstring and one usage example

DEPENDENCIES:
- Standard library only

EXAMPLE:
average([10, 20, 30]) -> 20.0

Edge Cases to Always Specify

  • Empty input, single item, negative numbers, zero division.
  • What to return on failure (None, error, exception?).
  • Performance constraints (large inputs?).
  • Mutating the input vs. returning a copy.

Ask for Tests, Then Verify

The professional workflow:

1. "Write the function per the spec."
2. "Now write 5 test cases including edge cases."
3. "Run the tests and fix any failures."

Never trust generated code blindly — run it, review it, and test edge cases yourself. AI code can be subtly wrong (off-by-one, wrong library API, security holes).

Common Code Prompt Mistakes

Mistake Fix
No signature/return spec Define the exact interface
No example Show input → output
No constraints State allowed libraries and limits
No edge cases Ask for edge-case handling explicitly
No tests Chain a test-generation step

Example: Refactor Request

Refactor this function to be faster and easier to test:
[code]

Keep the same public interface (same name, params, return).
Use only the standard library. Add a short comment explaining
the new approach.

Key Takeaways

  • Code prompts need a contract: context, output, dependencies, examples (CODE).
  • Specify edge cases explicitly and ask for tests.
  • Verify generated code by running it — never trust it blindly.
  • Keep the public interface stable when asking for refactors.

Next up: Prompting for writing and content — tone, audience, and the WRITE framework.

python · worked example
# The CODE framework: Context, Output, Dependencies, Examples
prompt = """Write a Python function that computes the average of a list of numbers.

CONTEXT:
- Input: a list of floats, e.g. [10, 20, 30]
- Handle empty lists (return None)

OUTPUT:
- A function named average(values)
- Include a docstring and one usage example

DEPENDENCIES:
- Standard library only

EXAMPLE:
average([10, 20, 30]) -> 20.0
"""

print(prompt)

# The generated code - verify it actually works
def average(values):
    """Return the mean of a list of numbers, or None if empty."""
    if not values:
        return None
    return sum(values) / len(values)

print("Result:", average([10, 20, 30]))
print("Empty list:", average([]))
Read-only sample

Lesson Code (python)

# The CODE framework: Context, Output, Dependencies, Examples
prompt = """Write a Python function that computes the average of a list of numbers.

CONTEXT:
- Input: a list of floats, e.g. [10, 20, 30]
- Handle empty lists (return None)

OUTPUT:
- A function named average(values)
- Include a docstring and one usage example

DEPENDENCIES:
- Standard library only

EXAMPLE:
average([10, 20, 30]) -> 20.0
"""

print(prompt)

# The generated code - verify it actually works
def average(values):
    """Return the mean of a list of numbers, or None if empty."""
    if not values:
        return None
    return sum(values) / len(values)

print("Result:", average([10, 20, 30]))
print("Empty list:", average([]))

Console Output

Write a Python function that computes the average of a list of numbers.

CONTEXT:
- Input: a list of floats, e.g. [10, 20, 30]
- Handle empty lists (return None)

OUTPUT:
- A function named average(values)
- Include a docstring and one usage example

DEPENDENCIES:
- Standard library only

EXAMPLE:
average([10, 20, 30]) -> 20.0

Result: 20.0
Empty list: None

Code Visualization Tips

  • Draw the CODE framework as a 4-box spec sheet the model must fill in.
  • Show bad vs. good code prompts side by side, annotating what each added line fixes.
  • Sketch the verify loop: generate → test → fix → re-test until green.

Professional Tips & Tricks

  • Include one runnable example pair — models infer intent from examples better than prose.
  • Ask for tests FIRST, then code — the tests act as a spec.
  • For refactors, paste the current interface and forbid changing it.

Python Code Judge & Practice Arena

LeetCode Style

Run real Python 3.12 WebAssembly code directly in your browser against automated test suites.

Solved:0 / 1
0 / 20 XP
Challenges:
Problem 1 of 1

Write a CODE Prompt

Medium+20 XP
Write a CODE-framework prompt for a function that counts word frequency in a string, including edge cases (punctuation, case) and one example.
main.pyPython 3.12 (WASM)
1
2
3
4
5
6
7
8
9
10
11
12
Press Run Code to test or Submit to verify test cases

Test Your Knowledge

Instant feedback

Quick Check: Prompting for Code

1 / 2
What does the CODE framework stand for?

Up next · Continue learning

Prompting for Writing & Content

The WRITE framework for audience, tone, and format — with before/after examples for posts, emails, and articles.

10 mins read50 mins
Start next lesson
Previous: Prompt Chaining & Multi-Step WorkflowsNext: Prompting for Writing & Content
Made withbyAmol Shukla·amolshukla.online
ASAmol Shukla

AI Developer, Trainer & Agentic AI Expert building practical learning systems and real-world AI applications.

Explore

  • Projects
  • Learning Hub
  • Prompts
  • Skills
  • Contact
  • Experience
  • Blogs

Connect

  • Resume
  • Contact
© 2026 Amol Shukla·Created withbyamolshukla.online
Back to top