Skip to content

A Claude Code / Codex plugin that helps you learn how to build while AI writes the code.

Python

feat: give learning checkpoints stronger visual separation

Noah Kimdd216d3b3102
diff --git a/.claude-plugin/plugin.json b/.claude-plugin/plugin.json
index 69cd03d..8676b90 100644
--- a/.claude-plugin/plugin.json
+++ b/.claude-plugin/plugin.json
@@ -1,6 +1,6 @@
{
"name": "sensible-vibes",
- "version": "0.1.6",
+ "version": "0.1.7",
"description": "You build. AI writes. Learn how to build software while Claude writes the code.",
"author": { "name": "Noah Kim" },
"repository": "https://github.com/nykooi1/sensible-vibes",
diff --git a/README.md b/README.md
index 9c35a36..650299e 100644
--- a/README.md
+++ b/README.md
@@ -28,15 +28,21 @@ Setup asks one question at a time. Use the arrow keys and Enter for choices; pic
```text
You: Add Stripe subscriptions.
-Claude: ✦ BUILD CHECKPOINT - Handling duplicate payment events
+Claude:
+────────────────────────────────────────
+✦ BUILD CHECKPOINT - Handling duplicate payment events
+
Stripe can deliver the same event more than once.
+
How would you prevent it from updating a subscription twice?
You: Save the event ID and check whether we already handled it?
Claude: Yes—that's the idea behind idempotency.
+────────────────────────────────────────
✦ DECISION REVIEW - Handling duplicate payment events
+
We'll save the event ID and update the subscription in one
transaction, so concurrent deliveries can't process it twice.
Next I'll implement that check in the webhook handler.
diff --git a/docs/development.md b/docs/development.md
index 2750ca4..418669c 100644
--- a/docs/development.md
+++ b/docs/development.md
@@ -66,6 +66,9 @@ Launch with `claude --plugin-dir /absolute/path/to/sensible-vibes`.
8. **Diagrams:** During orientation or a system check, confirm a compact terminal
diagram shows real components and labeled flows. Unknowns and proposals must
stay explicit; diagrams before reasoning must not silently decide the solution.
+ Checkpoints and reviews should start with a divider and bold named title,
+ followed by brief context and a bold question (or native picker), with no
+ trailing paragraphs obscuring the point where the learner should respond.
Do not commit `.sensible-vibes/` or test transcripts. The plugin recommends an
ignore rule during onboarding, but changes `.gitignore` only after telling the
diff --git a/skills/learn/behavior.md b/skills/learn/behavior.md
index ba7cce7..b51a6c5 100644
--- a/skills/learn/behavior.md
+++ b/skills/learn/behavior.md
@@ -21,8 +21,8 @@ dependency order, one learning loop at a time. Don't silently choose a stack or
storage model, or treat a feature requirement as a settled technical decision.
Keep undecided choices visible in the project map.
-1. **Orient.** Briefly explain the problem and any unfamiliar terms. Don't give
- away the solution before asking them to think.
+1. **Orient.** Open the named checkpoint, then briefly explain the problem and
+ any unfamiliar terms. Don't give away the solution before asking them to think.
2. **Ask.** Label every reasoning prompt, including follow-ups, with
`✦ BUILD CHECKPOINT - <specific decision name>` and ask one question
about how they'd approach it and why. Wait for their answer before proposing
@@ -61,8 +61,12 @@ with one question, 2–4 options, a short header (at most 12 characters), and
`multiSelect: false`. Only use a text menu when the tool is unavailable. Open-ended
reasoning uses one chat question. Every marked callout uses `✦ <TYPE> - <description>`;
never use a bare label, including for system checks or explanatory callouts.
-Use simple named titles; no decorative boxes,
-simulated animation, or shell commands to draw UI.
+Give each callout its own visual space: a Markdown divider (`---`), a blank line,
+then the **bold named title**. Put brief context beneath the title, not paragraphs
+ahead of it. Add a small diagram if useful, leave a blank line, then **bold the
+single question**. When awaiting input, end there (or show the native picker);
+don't bury the pause under more paragraphs. No decorative boxes, simulated
+animation, or shell commands to draw UI.
Follow the profile: Light = major decisions; Normal = important decisions;
Frequent = smaller meaningful decisions. Defaults are Normal, open-ended reasoning,