Mokelay Server

Mokelay public API service. It owns Mokelay orchestration execution and the PostgreSQL/Drizzle schema that used to live in mokelay-website.

API

• GET|POST /api/mokelay/{API_JSON_UUID}
• GET /api/database/schema
• POST /api/mokelay/analyze-data-source
• POST /api/mokelay/ai-generate-dsl
• POST /api/mokelay/ai-translate

Auth-like flows such as register, login, current user, and logout are exposed through Mokelay orchestration JSON definitions under server/assets/mokelay-apis and use the internal signed mokelay_orchestration_session HTTP-only cookie. Runtime loading checks local server assets first, then Cloudflare R2, then published records in the apis table. See docs/auth-json-apis.md for the generated interface documentation. In production, set COOKIE_DOMAIN=.mokelay.com so www.mokelay.com can call api.mokelay.com with credentials.

OAuth login and signup also use Mokelay orchestration JSON definitions. The public entrypoints are GET /api/mokelay/oauth_google_start, GET /api/mokelay/oauth_google_callback, GET /api/mokelay/oauth_github_start, and GET /api/mokelay/oauth_github_callback. Set OAUTH_CALLBACK_BASE_URL to the public API origin, for example https://api.mokelay.com.

Pages and API Builder outputs are exposed through Mokelay orchestration JSON definitions under server/assets/mokelay-apis, Cloudflare R2 object keys under mokelay-apis/*.json, or published apis table records. See docs/api-json-schema.md for the full API JSON schema and docs/orchestration-blocks.md for block configuration. Database blocks read connections from ${datasource}_DATABASE_URL, based on each block's inputs.datasource.

Read public database table metadata with GET /api/database/schema. The response is { "tables": [{ "name": "users", "columns": [{ "name": "id", "type": "uuid", "dataType": "uuid" }] }] }.

POST /api/mokelay/analyze-data-source recognizes data sources from an image or text through the generic OpenAI JSON block. Send image files as multipart data in the image field as JPEG, PNG, or WebP up to 10MB:

curl -X POST http://127.0.0.1:8787/api/mokelay/analyze-data-source \
  -F "image=@./source.png"

Send text as JSON with a non-empty userInput field:

curl -X POST http://127.0.0.1:8787/api/mokelay/analyze-data-source \
  -H "Content-Type: application/json" \
  -d '{ "userInput": "GET https://api.mokelay.com/api/mokelay/me?debug=true" }'

When the input contains JSON data, data is { "type": "JSON", "rawData": ... }. When it contains API information, data is { "type": "API", "domain": "...", "path": "...", "method": "...", "headerData": [], "bodyData": [], "queryData": [] }. Unrecognized input returns { "type": "UNKNOWN" }.

POST /api/mokelay/ai-generate-dsl generates Mokelay page DSL, API DSL, and capability upgrade specifications from a requirement document. See docs/ai-dsl-generation.md for the response contract.

curl -X POST http://127.0.0.1:8787/api/mokelay/ai-generate-dsl \
  -H "Content-Type: application/json" \
  -d '{ "requirementDocument": "客户管理：需要客户列表、创建客户、删除客户，删除前需要确认。" }'

POST /api/mokelay/ai-translate translates an ordered string array:

curl -X POST http://127.0.0.1:8787/api/mokelay/ai-translate \
  -H "Content-Type: application/json" \
  -d '{ "texts": ["Hello", "Welcome, {{name}}"], "sourceLanguage": "English", "targetLanguage": "中文" }'

The response data is { "translations": ["你好", "欢迎，{{name}}"] } and preserves the input order.

Local Development

npm install
cp .env.example .env
npm run dev

The dev server listens on http://127.0.0.1:8787. Mokelay orchestration JSON examples use Mokelay_DATABASE_URL.

Mokelay API JSON on R2

Runtime loading checks server/assets/mokelay-apis and Nitro server assets first, then Cloudflare R2 when all R2 environment variables are configured, then published rows in the apis table. R2 object keys use mokelay-apis/{API_JSON_UUID}.json by default.

Required R2 environment variables:

CLOUDFLARE_R2_ACCOUNT_ID=your-cloudflare-account-id
CLOUDFLARE_R2_ACCESS_KEY_ID=your-r2-access-key-id
CLOUDFLARE_R2_SECRET_ACCESS_KEY=your-r2-secret-access-key
MOKELAY_APIS_R2_BUCKET=your-r2-bucket
MOKELAY_APIS_R2_PREFIX=mokelay-apis

CLOUDFLARE_R2_ENDPOINT is optional. When omitted, the server uses https://<CLOUDFLARE_R2_ACCOUNT_ID>.r2.cloudflarestorage.com.

Sync the repo JSON files to R2 after editing them:

npm run sync:mokelay-apis:r2

The sync command automatically loads .env from the mokelay-server directory.

The runtime token needs Cloudflare R2 Object Read permission for fallback loading and Object Write permission when API builder users publish APIs. The sync command also needs Object Read & Write permission.

Database

npm run db:generate
npm run db:migrate

Production should use the existing Mokelay database URL so user records and migrations are preserved.

Deployment

Deploy to Vercel and bind api.mokelay.com.

Required environment variables:

NODE_ENV=production
Mokelay_DATABASE_URL=your-production-postgres-url
SESSION_SECRET=use-a-strong-random-string-at-least-32-chars
COOKIE_DOMAIN=.mokelay.com
CORS_ORIGINS=https://www.mokelay.com,https://mokelay.com,https://editor.mokelay.com,http://localhost:5173
OPENAI_API_KEY=your-openai-api-key
OPENAI_MODEL=gpt-4.1-mini
OAUTH_CALLBACK_BASE_URL=https://api.mokelay.com
OAUTH_APP_BASE_URL=https://www.mokelay.com
OAUTH_GOOGLE_CLIENT_ID=your-google-oauth-client-id
OAUTH_GOOGLE_CLIENT_SECRET=your-google-oauth-client-secret
OAUTH_GITHUB_CLIENT_ID=your-github-oauth-client-id
OAUTH_GITHUB_CLIENT_SECRET=your-github-oauth-client-secret
CLOUDFLARE_R2_ACCOUNT_ID=your-cloudflare-account-id
CLOUDFLARE_R2_ACCESS_KEY_ID=your-r2-access-key-id
CLOUDFLARE_R2_SECRET_ACCESS_KEY=your-r2-secret-access-key
MOKELAY_APIS_R2_BUCKET=your-r2-bucket
MOKELAY_APIS_R2_PREFIX=mokelay-apis

Before production signups, run migrations against the production database:

Mokelay_DATABASE_URL="your-production-postgres-url" npm run db:migrate

Verification

npm run typecheck
npm run test
npm run build