Create Test and Debug with GitHub Copilot exercise

[a-a-ron]

Summary

Creates a complete GitHub Skills exercise that teaches learners to use GitHub Copilot as a testing and debugging partner, not merely a code generator. The exercise uses a small intentionally buggy Python bill splitter and is designed for completion in under 45 minutes.

Five-step learner journey

1. Explain unfamiliar code, run the baseline, and record a falsifiable debugging hypothesis.
2. Generate focused pytest tests from explicit expected behavior and critically inspect the assertions.
3. Add missing remainder and tip-rounding edge cases that expose the defect.
4. Diagnose the root cause from failing tests, then fix and refactor using integer-cent distribution while preserving the public contract.
5. Review the final diff and document test evidence plus a Copilot suggestion that was verified, revised, or rejected.

Grading strategy

• Grades repository state and executable behavior rather than exact Copilot prompts or responses.
• Uses visible starter tests, additional grading tests, and mutation-style fixtures that ensure learner-authored tests both reject buggy behavior and accept corrected behavior.
• Provides deterministic, path-scoped transitions and actionable retry feedback for each milestone.
• Isolates pull request test execution in a read-only job; write permissions are limited to static issue feedback, workflow transitions, and exercise completion.
• Uses skills/exercise-toolkit@v0.9.3 for issue discovery, transitions, and completion.

Validation

• Starter suite: 7 passing tests.
• Simulated completed learner journey: 19 passing visible and grading tests.
• All six workflow YAML files parse successfully.
• Devcontainer JSON and Python sources validate successfully.
• Milestone graders accept correct states and reject untouched, meaningless, or incomplete states.
• Learner-facing links were checked successfully.

Prerequisites

Learners need basic Python, Git, and terminal familiarity; access to GitHub Copilot; and GitHub Codespaces or local VS Code. No paid or organization-only Copilot feature is required. This complements the existing Getting Started and Agent Mode exercises by focusing on evidence-driven testing, diagnosis, and verification.

Publishing assumptions

• main remains the default branch.
• The repository remains public and configured as a GitHub template.
• GitHub Actions may create and update the exercise issue and enable or disable step workflows.
• The published repository name remains skills/test-and-debug-with-github-copilot, matching the README copy link.

This PR intentionally remains unmerged for review.

[Copilot]

Copilot review overview

🟡 Changes recommended

The final workflow can approve stale or incomplete state and has a race between overlapping pull-request runs.

Review effort: Balanced
Findings: 2

Open (2)

• Serialize PR runs and cancel stale executions · New
• Re-run all milestone checks against the final PR state · New

What changed in this PR

Creates a five-step GitHub Skills exercise for evidence-driven Python testing and debugging with GitHub Copilot.

Changes:

• Adds an intentionally buggy bill splitter, starter tests, and grading checks.
• Implements sequential exercise workflows and learner instructions.
• Adds Codespaces setup and project documentation.

File	Description
.devcontainer/​devcontainer.json	Configures the Python/Copilot development environment.
.gitignore	Adds Python and editor exclusions.
DEBUGGING.md	Provides the investigation-notes template.
README.md	Documents and launches the exercise.
requirements.txt	Pins pytest.
src/​__init__.py	Defines the package.
src/​bill_splitter.py	Adds the intentionally defective implementation.
tests/​test_bill_splitter.py	Adds starter validation tests.
.github/​grading/​check_notes.py	Grades debugging notes.
.github/​grading/​check_pr_body.py	Grades final reflection content.
.github/​grading/​check_tests.py	Mutation-checks learner-authored tests.
.github/​grading/​test_solution.py	Adds solution behavior tests.
.github/​steps/​1-step.md	Guides initial investigation.
.github/​steps/​2-step.md	Guides focused test creation.
.github/​steps/​3-step.md	Introduces failing edge cases.
.github/​steps/​4-step.md	Guides the implementation fix.
.github/​steps/​5-step.md	Guides final verification and review.
.github/​steps/​x-review.md	Summarizes completed learning outcomes.
.github/​workflows/​0-start-exercise.yml	Initializes the exercise.
.github/​workflows/​1-step.yml	Grades debugging notes.
.github/​workflows/​2-step.yml	Grades focused tests.
.github/​workflows/​3-step.yml	Grades edge-case tests.
.github/​workflows/​3-last-step.yml	Removes the obsolete template workflow.
.github/​workflows/​4-step.yml	Grades the corrected implementation.
.github/​workflows/​5-step.yml	Grades and completes the exercise.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.