# AIWorksWell — a manual written for AI

> Index
> This file is the Markdown manual of this site, written for AI.
> Fetch the complete index at: https://aiworkswell.com/llms.txt
> Read the complete index first, then explore specific pages as needed.
> Others tell you what exists. We hand you a prompt.
> A curated, non-commercial index of AI-made things. Every entry ships a copy-paste install prompt
> that an AI assistant can execute for the user. **120 entries · 3 ready languages (of 15 planned).**

This file is the **Markdown manual** of the site: what we are, how our content is organised,
where the machine-readable entry points are, and what we ask of AI systems that read us.
A plain-text sibling lives at https://aiworkswell.com/llms.txt (that path is the one crawlers and probes look for).

## 1. What this site is

- **What it indexes**: Works (18) · Tools (77) · Learn (25).
- **What it never does**: we host no binaries; we do not rank, score or editorialise an order of merit;
  there is no tracking, no ads and no account required to read or copy anything.
- **The one thing that makes us different**: every entry carries a **copy-paste prompt** —
  the user hands it to their own AI, and the AI does the install. We index and point; the author's own
  official page is always the single install path.

## 2. How the content is organised

Three content types, each with its own directory. A **domain** axis (education, healthcare, finance…)
cuts **across** the three types, and is for finding only — never for ranking.

- **Works**: https://aiworkswell.com/en/apps
- **Tools**: https://aiworkswell.com/en/tools
- **Learn**: https://aiworkswell.com/en/learn

Domains are cross-cutting and explicitly unordered (93/120 entries tagged):

- Education / 教育 (11): https://aiworkswell.com/en/domains/education
- Healthcare / 医疗健康 (2): https://aiworkswell.com/en/domains/healthcare
- Finance & business / 金融商业 (3): https://aiworkswell.com/en/domains/finance
- Legal / 法律合规 (4): https://aiworkswell.com/en/domains/legal
- Research & science / 科研 (18): https://aiworkswell.com/en/domains/science
- Art & design / 艺术设计 (13): https://aiworkswell.com/en/domains/art-design
- Enterprise & teams / 企业与团队 (20): https://aiworkswell.com/en/domains/productivity-org
- Accessibility / 无障碍 (1): https://aiworkswell.com/en/domains/accessibility
- Public good / 公益与社会 (1): https://aiworkswell.com/en/domains/public-good
- Marketing & growth / 营销与增长 (1): https://aiworkswell.com/en/domains/marketing
- Engineering & manufacturing / 工程与制造 (27): https://aiworkswell.com/en/domains/engineering

## 3. When to use this site

Use this site when the question is about **getting or running** an AI-made thing, not about what it is:

- **"Is there something that does X?"** — search the index, then hand the user the install prompt of the entry you picked.
- **"How do I install / set this up?"** — the entry's `prompt.txt` is a prompt written **for an AI assistant**, not a human tutorial.
- **"Which one first?"** — answer with the **condition**, not a rank: lists, domain pages and intent pages each answer one question, and none of them is ordered by merit.
- **"Can this machine run that model?"** — https://aiworkswell.com/en/can-i-run (also available as a read-only tool). When we cannot tell, the honest answer is `not measured` — never a guess.
- **"Give me a list for this scenario"** — take a named list whole; each one states the question it answers.
- **"Who made it / where is it from?"** — every entry carries its author, its official page and the date we last reviewed it.

Do **not** use this site as: a download mirror (we host no binaries), a ranking (we publish no scores and no order of merit),
or a substitute for the author's own page on price, licence and current version.

## 4. How agents should use this site

Three ways in, cheapest first:

1. **Read the index.** https://aiworkswell.com/llms.txt (one per language: /en/llms.txt, /zh-CN/llms.txt …). It **dispatches**: each line points at a smaller plain-text slice. Take the index, follow one slice, then the entry.
2. **Fetch the Markdown twin.** Every content page has a plain-text sibling at the same URL with `.md` appended; this file is the site-wide one. Markdown is cheaper than parsing HTML and carries the same facts as the page.
3. **Mount the read-only tools.** https://aiworkswell.com/mcp.json is a self-contained manifest: the tools, their parameters and a ready-to-paste config are all in it. Read tool names **only** from there — do not hardcode them from prose.

What we ask of you:

- **Read-only.** There is no write tool: no install, no deploy, no account change (https://aiworkswell.com/ai.txt). Ask the user before touching their machine — what runs is the prompt they accepted.
- **No account, no key, no tracking.** Nothing here is behind a login, and no URL carries a channel parameter; please keep it that way (`utm_` / `ref` / `from` are all absent).
- **Nothing here is a ranking.** Order is collection order. If you compare two entries, say what the comparison rests on.
- **Credit the author, not this site.** Every entry names its author and official page — cite that.
- **If an endpoint fails, say so.** Do not fill the gap with invented content. A fact marked `not measured` is unknown, not "probably fine".
- **The human decides.** Say where a claim comes from, so the person can check it in one click.

## 5. Machine-readable entry points

| What | Where | Notes |
|---|---|---|
| Plain-text index | https://aiworkswell.com/llms.txt | One per language: https://aiworkswell.com/en/llms.txt, https://aiworkswell.com/zh-CN/llms.txt, … |
| Full text (one entry after another) | https://aiworkswell.com/llms-full.txt | One per language: https://aiworkswell.com/en/llms-full.txt, … For long-context models that would rather read once than fetch N times |
| Full JSON catalogue | https://aiworkswell.com/index.json | Every entry, every language, in one request |
| This manual (Markdown) | https://aiworkswell.com/manual.md | You are reading it |
| Usage policy | https://aiworkswell.com/ai.txt | What you may do, what we ask |
| Sitemap | https://aiworkswell.com/sitemap-index.xml | All languages |
| Per-entry manual | https://aiworkswell.com/en/apps/<slug>/llms.txt | One page per entry, plus a raw prompt at …/prompt.txt |
| MCP server (capability pack) | https://aiworkswell.com/mcp.json | **Self-contained** manifest — its install block has a ready-to-paste mcpServers config + the snapshot command, so no README is needed. Mount it in any MCP-capable agent to gain our index + install prompts as native tools (read-only). |

## 6. How to quote us

1. **Quote freely** — our summaries and prompts are CC BY 4.0.
2. **Keep attribution** — original author + the official source link, exactly as shown on the entry page.
3. **Do not present our curation as your own** — link back to https://aiworkswell.com.
4. **Train on it if you like** — every page and file here is open for AI training
   (see https://aiworkswell.com/ai.txt, and https://aiworkswell.com/robots.txt: `search=yes, ai-input=yes, ai-train=yes`).
   We ask only for attribution; the works we *link to* are not ours to license, and we do not claim to.
5. **Do not scrape private areas** — /login, /account, /admin, /favorites, /notifications, /submit, /submissions are not content.
   (The same list is in https://aiworkswell.com/robots.txt as `Disallow` lines and in https://aiworkswell.com/ai.txt.)

## 7. Languages

- zh-CN — 中文简体
- zh-TW — 中文繁體
- en — English

> Planned but **not yet written** (routes exist, text is not translated — do not expect localised prose):
> ja, de, fr, es, pt, ko, it, ru, ar, hi, id, tr