> Source URL: /unit-3/resources/final-project-ai-assistance.policy
# AI Assistance Policy for Final Projects

You are permitted and encouraged to use AI coding agents for your final project. These tools are becoming standard in professional development, and learning to use them effectively is part of this course.

## What You Can Use Agents For

- **Planning:** Brainstorming project ideas, outlining features, structuring your codebase
- **Debugging:** Finding and fixing errors in your code
- **Boilerplate and setup:** Creating project structure, configuration files, imports
- **Front-end code:** HTML, CSS, JavaScript, templates
- **Library-specific code:** Flask routes and views, Pygame rendering, Pandas operations, API calls
- **Documentation:** README files, comments, docstrings
- **Learning:** Asking agents to explain concepts, show examples, or walk through approaches

For Checkpoint 1 specifically, agents are great for helping you brainstorm, scaffold the project, install dependencies, and explain how the starter code fits together.

## What You Should Write Yourself

The core business logic of your project should be your own work. This means the main algorithms, data processing, and decision-making code that makes your project unique. If an agent helps you with core logic, make sure you understand every line and could rewrite it from scratch.

**Rule of thumb:** If someone asks you "how does this part work?" during your demo, you should be able to explain it confidently without looking at the code.

For your first working slice, try to hand-write the core logic yourself or direct the agent very closely while you build it. If an agent drafts that logic for you, rewrite it, annotate it, or explain it in your own words before you keep it.

**Checkpoint 1 standard:** You should be able to explain your first working feature line by line.

## A Good Week 1 Workflow

1. Use an agent to brainstorm and refine your spec.
2. Use an agent to scaffold the project and get the environment running.
3. Build one tiny, real feature that shows the core loop of your project.
4. For that feature, keep the core logic under your control and make sure you understand every line.
5. Record in your journal what you wrote yourself and what the agent helped with.

## Your Responsibilities

- **Understand your code.** You will demo your project and answer questions about it. If you cannot explain how something works, that is a problem regardless of who wrote it.
- **Use your AGENTS.md file.** This file tells agents about your project and what they should and should not help with. Keep it updated as your project evolves.
- **Document agent use in your journal.** In your `project.journal.md`, note when you used an agent for something significant. This is not a penalty; it helps your instructor understand your workflow and create better guides for you.
- **Commit often.** Frequent commits make it easy to see how your project evolved and to roll back if an agent makes a mess.

## Tools You Can Use

Any coding agent from the [Agent Comparison Guide](./agents-comparison.guide.md) is permitted. You may use multiple agents for different tasks. Chat-based AI (ChatGPT, Claude, Gemini web versions) is also fine for brainstorming and explanations.


---

## Backlinks

The following sources link to this document:

- [AI Assistance Policy](/unit-3/labs/01-transition/01-transition.lab.llm.md)