byndr-dev
Enables LLM-driven data enrichment for EPLAN article databases, allowing proposals for missing translations and descriptions through a Model Context Protocol server.
README
byndr-dev
An open platform for EPLAN teams, built on Cloudflare — data enrichment and automations in one place.
⚠️ Work in progress — largely conceptual. The design below describes where this is going; only the Foundation (Stage 1) is implemented today. Interfaces and scope will change. See
ARCHITECTURE.mdfor the roadmap.
Why
EPLAN article databases are full of gaps: missing translations and descriptions. And useful events (a PDF export, a BOM change) never leave the desktop. byndr-dev tackles both:
- Data + Gym. Ingest a read-only snapshot of your article database, then let an LLM close text gaps (translations, descriptions) through an MCP server. The validators are deterministic and run on the server — the model proposes, the server decides. Proposals are reviewed like pull requests; once a human approves, the change is queued and a small local client writes it back into EPLAN.
- Automations. Route EPLAN events (PDF/BOM exports, project events) to where an engineering team actually looks — Microsoft Teams, email, a shared drive, or a plain webhook you control.
Principles
- The cloud never writes to EPLAN. It produces proposals and a write queue; a local client applies only human-approved changes.
- Overlay, never mutation. The gym's only output is proposals; it never edits ingested data.
- Your data stays yours. Real article data lives in a private database, never in this repository. You bring your own.
- Per-tenant from day one.
Quickstart (development)
npm install
npm run dev # local worker
npm test # run the test suite
Create a tenant (prints its API key once):
node scripts/create-tenant.mjs "My Team"
Self-hosting
byndr-dev runs on Cloudflare (Workers + D1 + Durable Objects); deploy it to your own
Cloudflare account with Wrangler (see Quickstart). For teams that cannot use the
cloud, a fully-local runtime via workerd / Miniflare — the same runtime the test
suite already uses — is on the roadmap. Note: Wrangler configuration uses TOML or
JSON(C), not YAML.
Contributing
Contributions welcome — please read CONTRIBUTING.md. In short: TypeScript strict,
no any, no null, tests green, Conventional Commits.
License
MIT.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。