PROMPTS.md — how I used AI on this project

Copy this template

To start your own PROMPTS.md, open Show the text to copy below and use the copy button at the top right of the block, then paste it into a new file called PROMPTS.md in your repository. On GitHub, use the Copy raw file button on the template's GitHub page.

Show the text to copy

# PROMPTS.md — how I used AI on this project

> This log is required in every project. It is how you show your process, and it is part of your grade.
> Keep it honest and short. You do not need to log every message, only the ones that mattered.
> This file is public: no keys, no personal information, no classmates' names. Replace every [square-bracket placeholder].

**Project:** [Project name] · **Tools used:** [for example: Gemini (Canvas tool), GitHub Copilot agent mode, Groq API]

## Log

<!-- Copy this block for each significant prompt or session. Newest at the bottom. -->

### [date] — [what I was trying to do]

- **Tool:** [assistant or agent, and mode]
- **Prompt** (copy it in, or paste a share link to the chat if the tool makes one):

  ```text
  Goal:
  Context:
  Constraints:
  Done when:
  ```

- **What happened:** [Did it work? What did the AI get wrong?]
- **What I changed or checked myself:** [for example: read the diff, fixed a variable name by hand, tested with an empty input]
- **Teach-back** (🔴 AI off; for entries that produced code, written before you commit it): [2–3 sentences in your own words: what changed, and why it works.]
- **Commit:** [commit hash or link]

## Goldfish test (weeks 3 and 6)

<!-- Paste ONLY your SPEC.md (or AGENTS.md) into a brand-new chat and ask it to build or explain the project. Log what happened, then revise the doc. -->

- **Date and doc tested:** [date] · [SPEC.md or AGENTS.md]
- **What the fresh chat asked or got wrong:** [questions it asked, assumptions it made, features it invented]
- **What I changed in the doc:** [sections you rewrote; commit link]

## AGENTS.md trial (capstone, week 6)

<!-- Run two small tasks with your AGENTS.md in place. Did the agent follow it? -->

| Task | Rule I was checking | Followed? | Evidence (what the agent did) |
|---|---|---|---|
| [task 1] | [e.g. "plan first and wait for approval"] | [yes / partly / no] | [what happened] |
| [task 2] | [e.g. "never edit files outside public/ and lib/"] | [yes / partly / no] | [what happened] |

- **One change I made to AGENTS.md because of this trial:** [the change, and why]

## Ownership statement

<!-- Required for every submission, and graded. What did YOU write, decide or change? Be specific. -->

- **I wrote:** [code, tests or spec sections I wrote by hand]
- **I decided:** [design choices, what to build and what to leave out, AI suggestions I rejected and why]
- **I changed or fixed:** [bugs I found myself, AI-written code I rewrote]

## The `Assisted-by:` line

When AI helped with a commit, end the commit message with a blank line and an `Assisted-by:` line naming the tool. You learn this in week 2, and it's required from week 6:

```text
Add weekly summary chart

Assisted-by: GitHub Copilot (agent mode)
```

## What I'd prompt differently next time

-

This log is required in every project. It is how you show your process, and it is part of your grade. Keep it honest and short. You do not need to log every message, only the ones that mattered. This file is public: no keys, no personal information, no classmates' names. Replace every [square-bracket placeholder].

Project: [Project name] · Tools used: [for example: Gemini (Canvas tool), GitHub Copilot agent mode, Groq API]

Log

[date] — [what I was trying to do]

• Tool: [assistant or agent, and mode]

• Prompt (copy it in, or paste a share link to the chat if the tool makes one):

  Goal:
  Context:
  Constraints:
  Done when:

• What happened: [Did it work? What did the AI get wrong?]

• What I changed or checked myself: [for example: read the diff, fixed a variable name by hand, tested with an empty input]

• Teach-back (🔴 AI off; for entries that produced code, written before you commit it): [2–3 sentences in your own words: what changed, and why it works.]

• Commit: [commit hash or link]

Goldfish test (weeks 3 and 6)

• Date and doc tested: [date] · [SPEC.md or AGENTS.md]
• What the fresh chat asked or got wrong: [questions it asked, assumptions it made, features it invented]
• What I changed in the doc: [sections you rewrote; commit link]

AGENTS.md trial (capstone, week 6)

Task	Rule I was checking	Followed?	Evidence (what the agent did)
[task 1]	[e.g. "plan first and wait for approval"]	[yes / partly / no]	[what happened]
[task 2]	[e.g. "never edit files outside public/ and lib/"]	[yes / partly / no]	[what happened]

• One change I made to AGENTS.md because of this trial: [the change, and why]

Ownership statement

• I wrote: [code, tests or spec sections I wrote by hand]
• I decided: [design choices, what to build and what to leave out, AI suggestions I rejected and why]
• I changed or fixed: [bugs I found myself, AI-written code I rewrote]

The Assisted-by: line

When AI helped with a commit, end the commit message with a blank line and an Assisted-by: line naming the tool. You learn this in week 2, and it's required from week 6:

Add weekly summary chart

Assisted-by: GitHub Copilot (agent mode)

What I'd prompt differently next time