NyxID

Open-Source Agent Connectivity Gateway

中文版

Install NyxID Skills

Copy this line. Paste it into your AI agent (Claude Code, Cursor, or any LLM). Done.

Read https://raw.githubusercontent.com/ChronoAIProject/NyxID/main/skills/nyxid/SKILL.md and install the NyxID skill following its setup instructions.

Your agent reads the skill definition, installs the NyxID CLI, and configures itself. You don’t touch config files, don’t write YAML, don’t manage API keys manually.

Or install the CLI directly:

bash -c "$(curl -fsSL https://raw.githubusercontent.com/ChronoAIProject/NyxID/main/skills/nyxid/tools/install.sh)"

What NyxID Does

Your AI agents can’t reach your services. Your internal APIs, localhost dev servers, and private infrastructure are invisible to cloud AI agents.

NyxID is the open-source Agent Connectivity Gateway. One gateway between your APIs and your AI agents. Built in Rust. Self-hosted.

Credential-Injecting Proxy

NyxID sits between your agents and your APIs as a reverse proxy. When an agent makes a request, NyxID auto-injects the correct credentials at the network layer. Your agents never see or store raw API keys.

NAT Traversal

Your internal APIs and dev servers can be reached by AI agents through SSH tunnels. No public endpoint, no firewall changes, no VPN.

Run nyxid node start to expose private services to your agents.

REST → MCP Auto-Wrap

Point NyxID at any REST API. NyxID generates an MCP Server that any MCP-compatible client can consume. Any API becomes callable by any agent, instantly.

How It Works

Your Services (localhost / internal APIs / any REST endpoint)
  ↓
NyxID Gateway
  ├── Credential Injection (reverse proxy auto-injects keys per request)
  ├── NAT Traversal (SSH tunnels — expose private services)
  ├── MCP Auto-Wrap (REST → MCP Server, auto-generated)
  └── Per-Agent Isolation (scoped tokens, agents can't see each other)
  ↓
Any AI Agent (Claude Code / Cursor / n8n / any MCP client)

Built-in Skill Management

NyxID comes with Ornn, the AI Skill Platform, as its default service. Install NyxID, and your agent immediately has access to a library of skills: create posters, build presentations, pull data, collaborate through Lark and Slack.

Skills sync across every device where NyxID is installed. Search by what you want to do, not by skill name.

Explore Ornn →

What Only NyxID Does

No existing solution combines all three:

  1. Credential injection at the network layer — agents never touch raw keys. Not storage (everyone does that). Injection.

  2. NAT traversal for private APIs — expose localhost to cloud AI agents through SSH tunnels. Cloudflare Tunnel does NAT but has no credential injection. NyxID does both.

  3. REST → MCP auto-wrap — point at any REST endpoint, get an MCP Server. With built-in credential injection.

Open source. Self-hosted. Built in Rust.

97M monthly MCP downloads. 8,000+ servers exposed. $82K lost in 48 hours.

Your AI agents can’t reach your private services. And even when they can, credentials are exposed.

GitHub Read the Dossier →