# AI use and course mentoring AI assistance in this course is intended to support understanding, debugging, and independent work. You remain the author of the reasoning, implementation, and wording submitted for assessment. Follow your official course and activity instructions, including any more specific limits on AI use. ## Help you can request - Explanations of concepts, unfamiliar terminology, error messages, tracebacks, test failures, and existing feedback. - Questions and bounded hints that help you choose and carry out the next step. - Feedback on your own draft, code, design, or test approach without an AI replacement for your work. - Clearly different practice examples when they do not reveal the assessed reasoning or become the assignment solution with superficial changes. - Help understanding documented setup steps and commands, including their purpose, prerequisites, and effects. You run commands in the default mentor workflow and share relevant results with private information removed. - Drafts of ordinary GitHub Issues or Discussions asking for help or reporting a problem, based on your actual observations. You can ask a basic conceptual question without first producing code or completing a troubleshooting checklist. Ask your instructor through the documented course channel when instructions are missing or a barrier persists. ## Work that remains yours The AI must not create, complete, repair, rewrite, or supply submission-ready parts of assessed worksheets or deliverables. This includes assessed source code, tests, analyses, designs, diagrams, pseudocode, docstrings, explanations, reflections, and evidence of testing. The boundary applies to file changes, chat replies, patches, downloadable artifacts, worked examples, and other output locations. A sequence of small hints must not assemble the solution. The central decisions and expression must remain yours, even when you are stuck or close to a deadline. For debugging, the mentor can explain the failure, point to relevant evidence, and suggest an investigation. You make the substantive correction. For writing, it can discuss purpose, completeness, and unclear reasoning without rewriting an assessed passage for you. Preserve original-wording requirements, including any activity rule that excludes AI-generated docstrings. The mentor must not invent requirements, citations, observations, test results, completed steps, grades, extensions, or permissions. It must not search for answer keys or another student's solution to complete your activity. ## Issues and Discussions For an ordinary support post, the AI may draft a title and complete post from facts you provide: the affected activity, your goal, relevant environment, actual steps, expected and observed behavior, and the help requested. Unknown details must stay unknown rather than being invented. Review the draft before sharing it. Remove credentials, student records, private feedback, personal details, and unnecessary assignment code. A support post must not expose a completed solution. Drafting does not authorize the AI to publish, send messages, or change a GitHub Issue or Discussion. If an activity assesses the substantive content of a post itself, the same student-authorship boundary applies to that assessed content. The mentor can explain the structure and review your draft. ## Using the Course Mentor In a supported VS Code session, select **Course Mentor**. Its configuration requests only reading and searching, with no file editing, terminal execution, MCP tools, or implementation handoffs. Verify the tools actually available in the selected client and session before relying on that restriction. The workspace defaults disable Copilot inline completions and next-edit suggestions while preserving chat. Custom instructions do not govern inline suggestions, so the two configurations serve different purposes. Reusable prompts under `.github/prompts/` help you understand an activity, debug an attempt, review your work, or draft a support post. Prompt-file support depends on the client; use their body text manually where necessary. The mentoring boundary also applies when no saved prompt is used. AI is optional unless your official course instructions explicitly say otherwise. Preserve documented non-AI support and accessibility alternatives. Follow any established acknowledgement or disclosure requirements; the AI must not write an assessed personal reflection or invent an AI-use history. ## Course-template authoring This source repository contains generic course and activity templates. `{{AI TODO: ...}}` markers are intentional authoring placeholders for creating course-specific repositories and files. They are not student tasks and do not authorize a student-facing mentor to fill assessed work. Course authors resolve these markers and supply verified activity context before publishing student-facing material. Unresolved context does not permit the mentor to invent outcomes, rubrics, or assignment solutions. Template maintenance and student mentoring are separate working contexts. ## Limits of repository configuration Instruction files and workspace settings guide supported clients; they cannot prove which tools a person used or guarantee every model response. Personal settings, other clients, and tools outside this repository may behave differently. The configured mentor should still provide useful, bounded help and preserve student ownership across the conversation.