agentleFS
phpstanphpstan/phpstan-src/.claude/skills/regression-test/SKILL.md

regression-test

Skill400 starsChanged 48 days agoMITView on GitHub

Add a regression test for an already-fixed PHPStan bug given a GitHub issue number

4 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions4 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 4

    1. peteonrails/voxtype★ 1.8ksame name
    2. cloudposse/atmos★ 1.4ksame name
    3. open-guji/luatex-cn★ 127same name
    4. VodouAI/OS★ 1same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository3 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

11 rules. 0 are common, 6 are distinctive, and 5 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use when the bug is about wrong inferred type, missing type narrowing, or incorrect type after control flow — i.e., PHPStan reports the …

    line 43
  2. Very specific:

    Use assertType() and optionally assertNativeType() to pin the correct types

    line 46
  3. Distinctive:

    Use when the bug is about a false positive (rule reported an error that shouldn't exist) or a false negative (rule failed to report an …

    line 94
  4. Distinctive:

    Find the relevant rule test case: Search for the rule's identifier or class name:

    line 96
  5. Distinctive:

    Use the reproducing code from the playground sample for the test file.

    line 157
  6. Very specific:

    Put it in the correct namespace Bug<number>;.

    line 158
  7. Very specific:

    Add use function PHPStan\Testing\assertType.

    line 159
  8. Very specific:

    If the playground code requires a minimum PHP version, add // lint >= 8.0 (or whichever version) on the first line.

    line 160
  9. Distinctive:

    Run the specific test to confirm it passes with the current (fixed) code:

    line 165
  10. Distinctive:

    If the test passes even with the fix reverted, the test is not correctly covering the bug — revisit the assertions or the reproducing code.

    line 188
  11. Very specific:

    If you are submitting a pull request, append the PR description with a link to the issue with Closes …

    line 203

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Adding a regression test for a fixed PHPStan bug

You are given a GitHub issue number. Your goal is to write a regression test that locks in the fix so the bug cannot silently resurface.

Step 1 — Gather context from the issue

Fetch the GitHub issue and its comments:

gh issue view <number> --repo phpstan/phpstan --json title,body,comments

In the comments, look for:

  1. phpstan-bot's commit link — this identifies the commit that fixed the bug.
  2. PHPStan Playground links — URLs matching https://phpstan.org/r/<UUID>. These contain the reproducing code and the analysis results before and after the fix.

For each playground link, fetch the sample data:

curl -s 'https://api.phpstan.org/sample?id=<UUID>'

The API returns JSON with:

  • code — the PHP code that was analysed
  • level — the PHPStan rule level used
  • config.strictRules, config.bleedingEdge, config.treatPhpDocTypesAsCertain — configuration flags
  • versionedErrors — array of {phpVersion, errors: [{line, message, identifier}]}

The playground sample gives you the reproducing code and tells you what errors PHPStan produces (or should no longer produce) at each line. Use this to determine the correct test type and expected outcomes.

Step 2 — Decide which kind of test to write

Type-inference test (NSRT)

Use when the bug is about wrong inferred type, missing type narrowing, or incorrect type after control flow — i.e., PHPStan reports the wrong type for an expression but no specific rule error is involved.

  • Create tests/PHPStan/Analyser/nsrt/bug-<number>.php
  • Use assertType() and optionally assertNativeType() to pin the correct types
  • The file is auto-discovered — no test-case class changes are needed

Example (tests/PHPStan/Analyser/nsrt/bug-12875.php):

<?php // lint >= 8.0

declare(strict_types = 1);

namespace Bug12875;

use function PHPStan\Testing\assertType;

interface HasFoo
{
	public function foo(): int;
}

interface HasBar
{
	public function bar(): int;
}

class HelloWorld
{
	/**
	 * @param "foo"|"bar" $method
	 * @param ($method is "foo" ? HasFoo : HasBar) $a
	 * @param ($method is "foo" ? HasFoo : HasBar) $b
	 */
	public function add(string $method, HasFoo|HasBar $a, HasFoo|HasBar $b): void
	{
		assertType('int', $a->{$method}());
		assertType('int', $b->{$method}());

		$addInArrow = fn () => assertType('int', $a->{$method}());

		$addInAnonymous = function () use ($a, $b, $method): void {
			assertType('int', $a->{$method}());
			assertType('int', $b->{$method}());
		};
	}
}

Rule test

Use when the bug is about a false positive (rule reported an error that shouldn't exist) or a false negative (rule failed to report an error that should exist).

  1. Find the relevant rule test case. Search for the rule's identifier or class name:
    grep -r 'identifier' tests/PHPStan/Rules/ --include='*Test.php'
    Or look at the error identifier from the playground output (e.g., argument.type points to a rule in Rules/Methods/ or Rules/Functions/).
  2. Create the test data file at tests/PHPStan/Rules/<Category>/data/bug-<number>.php.
  3. Add a test method in the corresponding *Test.php rule test case.

Example — test data (tests/PHPStan/Rules/Methods/data/bug-11470.php):

<?php declare(strict_types = 1);

namespace Bug11470;

use DateTimeImmutable;

interface HelloWorld
{
	public function sayHello(): dateTimeImmutable;
}

interface HelloWorld2
{
	public function sayHello(dateTimeImmutable $a): void;
}

interface HelloWorld3
{
	public function sayHello(): ?dateTimeImmutable;
}

Example — test method in ExistingClassesInTypehintsRuleTest.php:

public function testBug11470(): void
{
	$this->analyse([__DIR__ . '/data/bug-11470.php'], [
		[
			'Class DateTimeImmutable referenced with incorrect case: dateTimeImmutable.',
			9,
		],
		[
			'Class DateTimeImmutable referenced with incorrect case: dateTimeImmutable.',
			14,
		],
		[
			'Class DateTimeImmutable referenced with incorrect case: dateTimeImmutable.',
			19,
		],
	]);
}

For false-positive fixes the expected-errors array is usually empty [], meaning the code should analyse clean.

Step 3 — Write the test

  • Use the reproducing code from the playground sample for the test file. Keep it as close as possible to the original. Once you have a reproducing failing test with the same error as the issue is about, do not edit the tested code sample.
  • Put it in the correct namespace Bug<number>;.
  • Add use function PHPStan\Testing\assertType; for NSRT tests.
  • If the playground code requires a minimum PHP version, add // lint >= 8.0 (or whichever version) on the first line.
  • For rule tests, determine the expected errors from the playground versionedErrors. If the fix eliminated a false positive, expect []. If the fix added a missing error, list the expected [message, line] pairs.

Step 4 — Verify the test

Run the specific test to confirm it passes with the current (fixed) code:

# NSRT test — run the full NSRT suite or a filtered subset:
vendor/bin/phpunit tests/PHPStan/Analyser/NodeScopeResolverTest.php --filter 'bug-<number>'

# Rule test — run the specific test class:
vendor/bin/phpunit tests/PHPStan/Rules/<Category>/<TestClass>.php --filter testBug<number>

Step 5 — Confirm the test catches a regression

Temporarily revert the fixing commit and verify the test fails:

git stash
git revert --no-commit <fixing-commit-hash>
vendor/bin/phpunit <test-command-from-step-4>
# Expect failure
git checkout .
git stash pop

If the test passes even with the fix reverted, the test is not correctly covering the bug — revisit the assertions or the reproducing code.

Step 6 — Commit

Stage the new/modified test files and commit.

git add tests/
git commit -m "Add regression test for #<number>

Closes https://github.com/phpstan/phpstan/issues/<number>"

Step 7 - Adjust pull request description

If you are submitting a pull request, append the PR description with a link to the issue with Closes https://github.com/phpstan/phpstan/issues/<number> so the upstream issue is closed automatically.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.