EEO

An open community where brands define themselves for AI.

EEO stands for Everything Engine Optimization, the umbrella term this community uses for SEO, AEO, GEO and LLMO.

EEO Audit is the measurement side: it asks AI engines the questions real buyers ask, then grades how visible a brand is inside the answers, with every original answer inspectable. The brand registry is the data side: 1,542,575 machine-readable brand cards following the eeo.brand.v1 schema (Wikidata CC0 bulk harvest plus hand-collected entries). Accuracy is enforced by AI review: every entry passed multiple rounds of per-entry AI judgment (1.5 million candidates judged across five passes; 46,883 non-organizations such as single events, persons, places, list pages and web pages rejected; the verdict files are archived in this repository), on top of deterministic rule-based cleaning, sharded as JSONL under datasets/registry/, which brands claim through domain verification in a public pull request. The site generator is the output side: it turns any set of cards into a static brand site that carries llms.txt, JSON-LD and a robots.txt allowlisting AI crawlers.

The standard behind all three pieces, including a five-rule conduct pact for optimization providers, is docs/eeo-standard.md.

When a buyer asks an AI for a recommendation, the answer is assembled from what the model has read: third-party articles, reviews, listing pages. The brand's own statement rarely reaches the model in structured form, so whoever writes about a brand ends up defining it there. This community starts at the source: the brand states its facts in an open format, proves domain ownership, and any engine, tool or buyer can read the result for free.

• Brands: claim your card. Find your brand in datasets/, then follow the domain-verification walkthrough in platform/README.md. Not listed yet? Submit a card per CONTRIBUTING.md.
• Developers: PRs are welcome across the repo, all of it small and dependency-free by design. Entry points: eeo-local.html (single-file edition), server.cjs with web/ (self-hosted edition), cli.cjs, action/, mcp/, extension/, tools/.
• Taxonomy writers: each industry needs a trait library and a buyer-question bank in standards/taxonomy/. Acceptance rules are in standards/taxonomy/README.md; the easiest first contribution is adding an industry alias.
• Engine watchers: propose new engine presets for the detection catalog, or plug any OpenAI-compatible endpoint into the CLI, server and MCP server via config.json. How to propose: CONTRIBUTING.md.

Download eeo-local.html, open it in a browser, tick engines in the catalog (26 presets, 11 of them China-native), paste the matching API keys, fill the brand form, run. Keys stay in the browser's localStorage. The report carries an S to D grade, awareness and mention scores, a per-engine breakdown and every original answer, exportable as JSON or printable to PDF.

cp config.example.json config.json    # fill in your keys
node cli.cjs check 蜜雪冰城 --industry 餐饮
node cli.cjs batch brands.txt --csv result.csv
node cli.cjs dataset 白酒 --limit 10

Zero dependencies, Node 18+. node cli.cjs --help lists all commands, including scheduled retests (watch) and dataset-driven batches.

docker build -t eeo-community .
docker run -p 8080:80 eeo-community

The image generates the brand site from the curated seed datasets/brands-1k.json (kept small on purpose so the container stays light) and serves it at http://localhost:8080: one page per brand, plus llms.txt, JSON-LD and robots.txt for AI crawlers. No API keys are involved.

For the self-hosted audit server instead, see server.cjs and config.example.json; the web UI builds with cd web && npm install && npm run build, and a prebuilt copy ships in public/.

eeo-open/
├── eeo-local.html       single-file audit tool, 26 engine presets
├── cli.cjs              batch CLI: check / batch / dataset / watch
├── core.cjs             shared audit engine used by CLI, MCP server, Action
├── server.cjs           self-hosted web edition, zero dependency
├── web/                 React UI source for the self-hosted edition (Vite)
├── public/              prebuilt copy of the web UI
├── action/              GitHub Action: brand visibility audit inside CI
├── mcp/                 MCP server for Claude Desktop, Cursor, Windsurf
├── extension/           Chrome MV3 quick-check extension
├── datasets/            registry/ sharded JSONL cards (1,542,575) + brands-1k.json curated seed
├── platform/            claim verification and finalize scripts
├── .github/workflows/   claim-verify, claim-finalize, monthly self-audit
├── standards/taxonomy/  8 curated industry libraries plus auto-generated banks (GB/T 4754, emerging, brand-derived)
├── tools/               gen-site / gen-llms / gen-schema generators + registry pipeline (harvest / clean / review merge / index)
├── docs/                EEO standard, ecosystem survey, publishing stack notes
├── examples/            sample S and D reports, claimed brand card, claim file
├── watch/               runtime dir for scheduled monitors (kept empty)
├── Dockerfile           community instance image
├── config.example.json  engine and key template for CLI / server / MCP
└── CONTRIBUTING.md      how to contribute cards, code, taxonomy, presets

EEO（Everything Engine Optimization，全引擎优化）是本社区对 SEO / AEO / GEO / LLMO 的统称。

买家向 AI 咨询推荐时，AI 引用的是模型读过的第三方内容；品牌自己陈述的信息很少以结构化形式到达模型，谁在写品牌，谁就在替品牌定义。本社区从源头做起：品牌用开放格式陈述自己的事实，通过域名验证公开认领，任何引擎、工具与买家都能免费读取。规范全文见 docs/eeo-standard.md。

社区目前有四块可用成果：

品牌方认领流程见 platform/README.md；数据、代码、分类法与引擎预设的贡献规则见 CONTRIBUTING.md。

Code is released under the Apache License 2.0. The standard text in docs/eeo-standard.md and the taxonomy files in standards/taxonomy/ are CC BY 4.0. Dataset cards: Wikidata-sourced cards are CC0, manually collected cards are Apache-2.0; the split is documented in datasets/README.md.

Copyright 2026 信阳市浉河区清白软件工作室 (Xinyang Shihe Qingbai Software Studio)