Editor-selected existing cases and prompts, separate from AI news. Selection dates use Hong Kong time.
2026-10-23 recheck: 5 existing, untested cases; last two are demos/reports. ChatGPT web/mobile, Claude web, Notebook web: TW conditional; HK/MO unlisted. Free limits; Copilot regions unknown. Notebook docs: Gemini Notebook; no mobile mind maps. TC quality, local paid prices, APIs unknown. Fictional student data only; no bookings, orders or code edits.
Keep a sanitized itinerary and preferences in one ChatGPT project, then discuss revisions and draft a list of changes and timing conflicts.
Prompt
Using this project’s itinerary, adjust [date/day and time slot] to meet [new activity requirements]. Change only what is necessary and preserve confirmed accommodation and transport. Flag conflicts with those arrangements instead of changing them. List before-and-after changes, possible conflicts and details still needing verification. Propose a draft itinerary only; do not cancel, change or make bookings.
Evidence limits and cautions
- Remove passport details, payment data, booking codes and private contact information.
- Project information does not automatically update when a travel operator publishes changes.
Original source: OpenAI：Projects in ChatGPT
Add a report to NotebookLM and generate an expandable map of its topics and related material to help choose what to read closely.
Prompt
For [branch name] in this mind map, identify the core concepts needed before reading. Explain them using the selected sources with citations, and flag insufficient source support. Explain whether its relationship with [other branch] is explicitly supported. Do not treat a conceptual connection as causation.
Evidence limits and cautions
- A connection between concepts does not necessarily imply causation.
- Feature names, devices and interfaces may change between versions.
Original source: Google：Use Mind Maps in Gemini Notebook
Give Copilot entry files and relevant modules to draft a file-referenced call-flow guide and open questions, then trace the actual code to verify it.
Prompt
Do not modify files yet. Starting from [entry file], trace the data flow for [feature]: inputs, key functions, outputs and side effects, with file locations. Separate directly observable facts from inferences. Finish with the shortest useful reading route.
Evidence limits and cautions
- AI may miss dynamic calls, configuration or runtime differences.
- Verify architecture explanations against code and tests.
Original source: GitHub Blog：Documenting and explaining legacy code
Supply your use, budget and must-haves to draft reusable buying criteria and a sourced comparison for your own decision.
Prompt
Help create reusable buying criteria for [product purpose], in [region], with [budget including necessary costs and currency], [must-haves] and [negotiable preferences]. Identify conflicts and gaps before researching matching products. Support comparisons with specifications and sources, noting unknowns, check times and tradeoffs. Do not choose material preferences or a fixed number of candidates for me. Do not order anything.
Evidence limits and cautions
This is a product-research demonstration, not durability testing or a quality guarantee. It cannot establish that an item will last a lifetime.
The following is an original, untested exercise, not a reproduction of the reported session. Check available features and permissions, then adapt it to your own constraints.
- Brand age, natural materials or advertising volume alone do not establish quality. Do not inherit the demonstrator’s tastes.
- Stored prices and availability age; recheck them before purchase.
Original source: Build a 'Buy It For Life' AI Shopping Assistant with Claude
Use entirely fictional observations to draft a practice comment with a note reference for each claim, without inventing traits or family context.
Prompt
Turn [entirely fictional teacher observations/tone/format/length] into a practice comment. Map each statement to a note. Add no personality, ability or family-background inferences; flag uncertainty. Practice faithful wording only. Do not grade, process real student data or send anything.
Evidence limits and cautions
Personal satisfaction does not establish comment accuracy; the model is not a student assessor.
These steps and prompts are original, untested exercises by this site, not the author’s verbatim workflow. Reported results do not guarantee exercise outcomes.
- Do not supply real children’s information, grades or health records.
- Do not make grading, admissions, disciplinary or other educational decisions.
Original source: DetailNo9969: teacher administration account