Test Generation

What the source does

• FilesystemNone observedNo reads or writes in the source
• NetworkNone observedNo outbound calls in the source
• ShellNone observedNo commands run in the source
• DependenciesPinnedNothing floats
• SecretsNone foundNo keys in source

Five audit layers

Gates applied: no behavioural pass, no license.

Audit 0.4.1 · source sha dd4da722e22a · 18 Oct 2026Report an issue or request a re-scan

• Publisheragulli GitHub userOwner of the repository this skill is read from
• Source folderagulli/atlas-agents/ch09_agent_skills/skills/test-generationThe exact folder we audited
• LicenceNot declaredAs the repository states it
• Commit pinned for this verdictdd4da722e22aA new audit runs when the source changes
• Repository statusNot archivedLast push 93 days ago
• Sandboxed runNot runSkills are read, not executed; layer L4 is skipped

Its description: Generate comprehensive test suites for existing code. Use when asked to write tests, add test coverage, create unit tests, or build integration tests.

---
name: test-generation
description: Generate comprehensive test suites for existing code. Use when asked to write tests, add test coverage, create unit tests, or build integration tests.
license: MIT
compatibility: Requires pytest (Python) or vitest/jest (Node.js)
---

## Process

1. **Read the source code.** Load every function and class under test. Do not write tests from a summary or description — you must see the implementation.

2. **Identify the test framework.** Check for:
   - `conftest.py` or `pytest.ini` → pytest
   - `vitest.config.*` → Vitest
   - `jest.config.*` → Jest
   If none are found, default to pytest for Python, Vitest for TypeScript/JavaScript.

3. **Categorize test types needed:**
   - **Happy path:** Normal inputs, expected outputs
   - **Edge cases:** Empty inputs, None/null, boundary values, maximum lengths
   - **Error cases:** Invalid inputs that should raise exceptions or return errors
   - **Integration:** Tests that verify multiple components work together

4. **Write tests following the AAA pattern:**
   ```python
   def test_descriptive_name():
       # Arrange — set up inputs and expected state
       user = create_test_user(name="Alice")
       
       # Act — call the function under test
       result = get_user_greeting(user)
       
       # Assert — verify the output
       assert result == "Hello, Alice!"
   ```

5. **Name tests descriptively.** The test name should describe the scenario and expected outcome:
   - ✅ `test_login_rejects_empty_password`
   - ❌ `test_login_2`

6. **Run the tests.** Execute the full test suite and confirm all tests pass. If any fail, fix the test (not the source code) unless you discover an actual bug.

## Rationalizations

| Excuse | Rebuttal |
|--------|----------|
| "I'll just write happy path tests" | Edge cases are where bugs live. Write at least one edge case per function. |
| "The function is too simple to test" | Simple functions get refactored. Tests protect against regressions during refactoring. |
| "I'll add tests later" | You won't. Write them now. |
| "Mocking everything is fine" | Over-mocking hides integration bugs. Mock external services, not your own code. |

## Verification

- [ ] Tests were executed (not just written)
- [ ] All tests pass
- [ ] At least one edge case test exists per function
- [ ] Test names describe the scenario, not just "test_1", "test_2"

Excerpt, first 2,391 characters of SKILL.md, verbatim from the audited commit. It is text from someone else’s repository, quoted so you can judge it before your agent reads it.

git clone https://github.com/agulli/atlas-agents
git -C atlas-agents checkout dd4da722e22a

Works in

1. 2026-10-18
   ✓SAFE B 89 / 100

   first audit · source dd4da722e22a

Questions about Test Generation

• github.com/agulli/atlas-agents/tree/main/ch09_agent_skills/skill — the audited source at commit dd4da722e22a and every finding aboveObserved · 18 Oct 2026

Spotted something wrong? Report a correction · Methodology