Document exercise: one missing “not” reverses the rule
Compare fictional originals and converted text for changed negation, version conditions and defaults.
Execution environment and usage
AI / software
Gemini
Where to operate
Ordinary text chat in the Gemini consumer web app at gemini.google.com. This card uses manually supplied text only.
Getting started
- Prefer an adult personal Google account and check regional/account eligibility first. Do not buy a plan for this exercise.
- Paste the fictional fixture into ordinary chat, or substitute authorized de-identified short text. Connect no external services.
- Compare output with the human acceptance checks. Preserve errors, gaps and unknowns for review.
Setup and plan requirements
Official web documentation checked 2026-11-11 lists Hong Kong, Macau and Taiwan, plus Simplified, Traditional and Hong Kong Chinese and English. An adult personal account is our recommended exercise scope, not a claim that all child, school or work accounts qualify. Personal/school accounts generally require 13 or the local applicable age; work accounts have separate 18+, administrator and plan conditions. Child/Family Link documentation contains inconsistent wording and needs separate verification.
Prompt portability limits
Start with standard free access; no subscription is required for this card. Taiwan advertises NT$0/month for free access. Local Hong Kong/Macau paid prices are unverified, so no local paid price is promised. Compute limits vary with model, complexity and conversation length, including five-hour and weekly caps; no fixed number of prompts is guaranteed. Wait if limited rather than assuming an upgrade is needed. Mobile, API, CLI and the original automation have separate eligibility, account and cost requirements.
Usage limitations
Documentation review is not a test from the three regions. Prompts, account access, model output and the original full workflow were not tested. Use fictional or de-identified data, never confidential inputs; some data can receive human review. Disabling activity is not zero retention: relevant chats can remain for 72 hours and undergo safety review. A person must verify outputs; text advice is not execution.
Environment sources
- https://support.google.com/gemini/answer/13575153?hl=en · 2026-11-11
- https://support.google.com/gemini/answer/13278668?hl=en · 2026-11-11
- https://support.google.com/gemini/answer/16275805?hl=en · 2026-11-11
- https://gemini.google/tw/subscriptions/?hl=zh-TW · 2026-11-11
- https://support.google.com/gemini/answer/13594961?hl=en · 2026-11-11
- https://hkfrees.com/ai/case/reported-docs-extraction-fidelity/ · 2026-11-11
- https://dev.to/sizzlebop/i-gave-my-ai-agents-their-own-documentation-crawler-and-pulled-60-pages-of-clean-markdown-in-49-2cl7 · 2026-11-11
適用情境 / When to use
Compare fictional originals and converted text for changed negation, version conditions and defaults. Adds known-answer logical reversals and invented-content checks, beyond file existence or word counts. Fictional fixture, not a real usage record: S1: Without approval, cache deletion is prohibited. S2: In v2, timeout defaults to 0, meaning no timeout. S3: Only v3 supports retry. Converted C1: Approval permits cache deletion. C2: timeout defaults to timing out after 0 seconds. C3: retry is supported. C4: Automatic backup is supported, with no matching original.
步驟 / Steps
- Preserve input conditions and existing labels and use the fictional fixture with the parameterized prompt in text chat.
- Human check: C1 loses S1’s prohibition and implies approval alone is sufficient; restore S1. C2 changes no timeout into immediate timeout and omits v2. C3 loses the v3 restriction. C4 is unsupported. These are not formatting differences, and no real product behavior was validated.
- If output fails a check, record the discrepancy and revise the request for human review. Never present an expected answer as observed model output.
提示詞 / Prompt
Compare [numbered original] with [numbered conversion] against [required semantics and version fields]. Separate omissions, meaning changes, unsupported additions and harmless formatting. Preserve negation, conditions, units and scope. Cite both line IDs and propose minimal textual repairs; invent no source content. Review supplied text only, without crawling pages or validating software behavior.
Fictional fixture, not a real usage record: S1: Without approval, cache deletion is prohibited. S2: In v2, timeout defaults to 0, meaning no timeout. S3: Only v3 supports retry. Converted C1: Approval permits cache deletion. C2: timeout defaults to timing out after 0 seconds. C3: retry is supported. C4: Automatic backup is supported, with no matching original.
留意事項 / Cautions
- Platform and region: ordinary Gemini web text chat. Official lists include Hong Kong, Macau, Taiwan, Chinese variants and English. Documentation checked 2026-11-11; local connections were not tested.
- Plan and account: prefer an adult personal account. Standard free access is a starting point, with variable usage caps and no required subscription. Taiwan advertises NT$0/month free access; Hong Kong/Macau paid prices are unverified. No particular model or prompt count is guaranteed. Work, school and child accounts need separate checks.
- No original toolchain is connected. Web availability does not establish mobile, API, CLI or original-automation access. Never supply confidential data. Activity-off chats can still remain for 72 hours and receive safety review.
- Use fictional documentation, not internal APIs, keys or restricted documents.
來源限制 / Evidence limits
The author describes local Pulpie MCP extraction and one timed 60-page crawl. This is self-report, not a fidelity or speed guarantee. This card compares supplied short text without crawling, installation or original-tool validation. Source published: 2026-10-05; checked: 2026-11-11. The original source provides inspiration, not validation of this exercise. Its toolchain, regional services and plans have not been reproduced. This card uses only the documented Gemini web text environment.
流程說明 / Workflow note
Original editorial adaptation, not a newly discovered use case or the source author’s verbatim workflow. Source-reported outcomes do not guarantee this exercise. Prompts were not tested. Adapted from this existing case: https://hkfrees.com/ai/case/reported-docs-extraction-fidelity/?lang=en Adds known-answer logical reversals and invented-content checks, beyond file existence or word counts. Open this card’s title for the official environment-source links and human acceptance steps on its full case page.
原始工具組合 / Original tool stack
Claude; pulpie-mcp; Markdown
Source publication date: 2026-10-05
Source checked: 2026-11-11