Vibe Coding — Prompting and Context flashcards
10 of 62 cards free to study here · 5 topics · the whole deck is built into Willys
Writing a spec with goal, users, constraints and done-when, giving examples and exact error messages, plan-first prompting, asking for tests, project rules files, keeping context focused, reviewing diffs and resetting when the AI is stuck.
Learn deck · part of Vibe coding · written against: Building software with AI assistants: workflow, prompting, git, web basics, debugging and security; tool-neutral. Updated 2026-10-10.
Study the free sample
10 cards from the deck as they are. Flip, rate yourself, filter by topic; your progress stays in this browser.
Tap the card or press Space to flip
The 10 sample cards as a list
-
Which four parts make a useful mini-spec for a feature you want built?Goal, users, constraints and "done when": what it should achieve, who uses it, what limits apply, and how you'll know it works.Writing a Spec › Goal and Users
-
Why state the goal of a feature, not just the steps you imagine?Knowing the why lets the AI choose a sensible approach, and notice when your suggested steps wouldn't actually get there.Writing a Spec › Goal and Users
-
Why does an example input with its expected output help so much?An example is concrete: it pins down the exact format, and any edge cases it includes, in a way a description often can't.Giving Good Input › Examples
-
You want dates shown in one particular style. What's the fastest way to get it right?Show an example of the finished format, such as "Tue 4 Mar, 14:05", instead of describing it in words.Giving Good Input › Examples
-
What is plan-first prompting?Asking the AI to outline its approach before writing code (which files it will change and how) so you can correct the plan while that's still cheap.Plans Tests and Rules › Plan First
-
Why is fixing a bad plan cheaper than fixing bad code?A plan is a few lines to read and edit. Code built on the wrong plan may have to be thrown away entirely.Plans Tests and Rules › Plan First
-
What is the context in an AI coding session?Everything the model can see right now: the conversation, your instructions and any files or output it has read. It has a size limit.Managing Context › Focused Context
-
Why can a very long conversation make the assistant's work worse?Old, irrelevant or contradictory details crowd the context, so the model may follow stale instructions or lose track of what matters now.Managing Context › Focused Context
-
After the AI edits code, why review the diff instead of rereading whole files?The diff shows only what changed, so you can check every edit, including surprise deletions, without wading through untouched code.Reviewing AI Output › Reviewing Diffs
-
Which red flags in a diff should make you pause?Code deleted that you didn't ask to remove, new dependencies, changed tests, hard-coded secrets and edits to unrelated files.Reviewing AI Output › Reviewing Diffs
Unlock the whole deck
You are studying a free 10-card sample. All 62 cards of Vibe Coding — Prompting and Context, the other 5 Vibe coding decks, and spaced-repetition scheduling are built into Willys. 14-day free trial, then $14.99 a year.
In Willys, tap Exam Library at the top of your Decks (or Browse Exam Library… under the + button), find the exam on its shelf or by search, tick it and tap Add.
More Vibe coding decks
← All Vibe coding decks, with the study guide and topic outline.
About these cards
Review status: Second-reader reviewed, with corrections applied.
Checked against:
- Original material; checked against Anthropic's Claude prompting guide, its engineering article on context engineering and the Claude Code documentation, OpenAI's prompting guides and Codex AGENTS.md documentation, GitHub Docs (CC BY 4.0), the Cursor Rules documentation, the OWASP Cheat Sheet Series (CC BY-SA 4.0), Stack Overflow's guide to minimal reproducible examples, Mozilla's bug-writing guidelines, Microsoft Learn's C/C++ build-errors page (CC BY 4.0), Wang et al. (2024) on deprecated API use in LLM code completion (CC BY 4.0) and the Git documentation; reference only, no text copied
These are independent study notes written for Willys. They are not official exam material, may contain mistakes, and are no substitute for your course, your teacher or the exam owner's own publications. Found an error? Tell us — corrections ship as a new version of the deck.
Use of these pages is covered by our Terms of Use.