How to use
- Copy the installation command and review the source package before installing it in your own agent environment.
- Check the source document for required tools, credentials and supported hosts.
- Describe your input and desired deliverable; review the output against the source material.
npx skills add https://github.com/caldoai/skills --skill prune-testsBefore you start
- Installation does not include a model subscription, API allowance or permission to access your files.
- Repository stars and installation counts describe activity, not an independent quality or security test.
Rendered from the source repository. Headings, examples, code, tables, links, and referenced images are preserved.
Prune every test that guards no behavior.
1. Find what each test depends on
Keep a test only if it depends on the intended behavior: it should change only when that behavior changes, and fail only when it regresses.
If another test already guards the same behavior, prune the copy.
Prune tests that depend on:
The past: it guards code that no longer exists, so it has no reason of its own to change.
For example:
- A feature was removed, and tests assert it's gone. Removed code never changes, so they can't catch a regression.
- A module was moved, and a test asserts it's absent from its old path.
Nothing: no change to the code can make it fail, so it has no reason of its own to change.
For example:
- It tests the browser: a
<button>renders, or a click event fires. The browser is already tested. - It mocks the unit under test, then asserts what the mock returns.
- It asserts what the type system already guarantees.
The implementation: it changes whenever the code does, even when behavior didn't.
For example:
- The expected value is derived from the same logic as the code, so a wrong result still passes.
- It asserts private state, internal call order, or which helper was invoked.
2. Prune
- Delete every test that isn't anchored to intended behavior.
- Where one also covers behavior something relies on, rewrite it against that behavior instead.
- Delete production code that only existed for the tests you removed.
What done looks like
- Every test left is anchored to intended behavior.
- Each one would fail on a regression, and only on a regression.
- No test asserts code that no longer exists.
- No two tests guard the same behavior.
- No production code exists only for tests.
Example
--- a/pricing.test.ts
+++ b/pricing.test.ts
@@ rewritten: the expected value repeated the code's logic @@
test("applies the discount", () => {
- const price = 100;
- expect(applyDiscount(price, 0.2)).toBe(price - price * 0.2);
+ expect(applyDiscount(100, 0.2)).toBe(80);
});
--- a/cart.test.ts
+++ b/cart.test.ts
@@ rewritten against what callers see: it asserted private state @@
test("adds an item to the cart", () => {
const cart = new Cart();
cart.add(book);
- expect(cart._items.length).toBe(1);
+ expect(cart.total()).toBe(book.price);
});
@@ deleted: checkout.test.ts already guards the same total @@
-test("total includes tax", () => {
- expect(cartTotal([book], { tax: 0.1 })).toBe(11);
-});
--- a/export.test.ts
+++ b/export.test.ts
@@ deleted: it guards code that no longer exists @@
-test("legacy exporter is gone", () => {
- expect(existsSync("src/export/legacy.ts")).toBe(false);
-});
--- a/session.test.ts
+++ b/session.test.ts
@@ deleted, with the production code that only existed for it @@
-test("clears the cache", () => {
- _resetCacheForTests();
- expect(_getCacheSize()).toBe(0);
-});
--- a/session.ts
+++ b/session.ts
@@ @@
-export function _resetCacheForTests() { cache.clear(); }
-export function _getCacheSize() { return cache.size; }