integration-hub
This skill should be used when the user asks to "IntegrationHub", "spoke", "flow action", "integration", "subflow", "connection alias", "credential alias", "REST step", or any ServiceNow IntegrationHub development.
SKILL.md
| Name | integration-hub |
| Description | This skill should be used when the user asks to "IntegrationHub", "spoke", "flow action", "integration", "subflow", "connection alias", "credential alias", "REST step", or any ServiceNow IntegrationHub development. |
Snow-Flow is an AI-powered, multi-agent development framework built specifically for ServiceNow. It combines a powerful terminal UI with 200+ ServiceNow MCP tools, 54 bundled domain skills, and support for 20+ AI providers to give you an autonomous coding agent that understands your ServiceNow instance.
<br> <table> <tr> <td width="33%" valign="top">ServiceNow-Native<br> <sub>200+ MCP tools and 54 domain skills purpose-built for ServiceNow development, deployment, and automation.</sub>
</td> <td width="33%" valign="top">Any AI Provider<br> <sub>Works with 20+ providers: Anthropic, OpenAI, Google, AWS Bedrock, Azure, Groq, and many more.</sub>
</td> <td width="33%" valign="top">Multi-Agent<br> <sub>Built-in build & plan agents with configurable permissions, custom agents, and subagent orchestration.</sub>
</td> </tr> <tr> <td width="33%" valign="top">Terminal-First<br> <sub>Beautiful TUI built with SolidJS + opentui.</sub>
</td> <td width="33%" valign="top">MCP Ecosystem<br> <sub>Connect any MCP server via stdio, SSE, or HTTP. Extend with the full MCP ecosystem.</sub>
</td> <td width="33%" valign="top">Open Source<br> <sub>Elastic License 2.0. Fully transparent. Community-driven.</sub>
</td> </tr> </table> <br>Getting Started
Install
curl -fsSL https://snow-flow.dev/install | bash
<details>
<summary><b>More installation methods</b></summary>
<br>
# npm / bun / pnpm / yarn
npm i -g snow-flow@latest
# Homebrew (macOS & Linux — recommended, always up to date)
brew install groeimetai/tap/snow-flow
# Windows
scoop install snow-flow
choco install snow-flow
# Arch Linux
paru -S snow-flow-bin
</details>[!TIP] Remove versions older than 0.1.x before installing.
Quick Start
snow-flow
On first launch, Snow-Flow prompts you to configure an AI provider. You can also pre-configure in snow-flow.jsonc:
{
"$schema": "https://snow-flow.dev/config.json",
"provider": {
"anthropic": {},
"openai": {},
},
}
<br>
Supported AI Providers
Snow-Flow is not coupled to any single AI provider. Use whichever model works best for you.
<table> <tr> <td><b>Anthropic</b></td> <td><b>OpenAI</b></td> <td><b>Google</b></td> <td><b>AWS Bedrock</b></td> </tr> <tr> <td><sub>Claude 4.5/4.6 Opus, Sonnet, Haiku</sub></td> <td><sub>GPT-5, GPT-4, o-series reasoning</sub></td> <td><sub>Gemini 2.5 Pro, Flash + Vertex AI</sub></td> <td><sub>All models, cross-region support</sub></td> </tr> <tr><td colspan="4"> </td></tr> <tr> <td><b>Azure</b></td> <td><b>GitHub Copilot</b></td> <td><b>Mistral</b></td> <td><b>Groq</b></td> </tr> <tr> <td><sub>OpenAI + Cognitive Services</sub></td> <td><sub>GPT-5 via Copilot enterprise</sub></td> <td><sub>Large, Medium, Small</sub></td> <td><sub>Ultra-fast inference</sub></td> </tr> <tr><td colspan="4"> </td></tr> <tr> <td><b>xAI</b></td> <td><b>OpenRouter</b></td> <td><b>GitLab AI</b></td> <td><b>+ 10 more</b></td> </tr> <tr> <td><sub>Grok models</sub></td> <td><sub>100+ models, single API</sub></td> <td><sub>Native GitLab integration</sub></td> <td><sub>Cohere, Perplexity, DeepInfra, Cerebras, Together AI, Vercel, ...</sub></td> </tr> </table> <br>Agents
Switch between agents with Tab:
| Agent | Description |
|---|---|
| build | Default agent with full tool access for development work |
| plan | Read-only agent for analysis and exploration — denies edits, asks before bash |
| general | Subagent for complex multi-step tasks — invoke with @general |
Custom agents can be configured in snow-flow.jsonc with per-agent model selection, permissions, and temperature controls. Learn more about agents.
Built-in Tools
19+ tools out of the box:
File Operations Shell Web Dev
─────────────── ───── ─── ───
read bash (streaming, webfetch plan (enter/exit)
write pty support) websearch task management
edit codesearch LSP (experimental)
glob skill invocation
grep
ls
apply_patch
<br>
ServiceNow MCP Integration
The core of Snow-Flow — 200+ MCP tools purpose-built for ServiceNow.
<table> <tr> <td width="33%" valign="top">Operations
- Query any table
- CMDB search
- User management
- Operational metrics
Development
- Script Includes
- Business Rules
- Client Scripts
- UI Policies & Actions
Automation
- Flow Designer
- Scheduled Jobs
- Approval Workflows
- Event Management
Deployment
- Widget deploy & preview
- Update Set lifecycle
- Artifact validation
- Rollback support
Security
- ACL management
- Domain Separation
- Compliance auditing
- Vulnerability scanning
Analysis
- Reporting & Dashboards
- KPI management
- Performance Analytics
- Data quality checks
Connect to your instance:
{
"mcp": {
"servicenow": {
"type": "local",
"command": ["snow-flow", "mcp", "start"],
"environment": {
"SERVICENOW_INSTANCE_URL": "https://your-instance.service-now.com",
"SERVICENOW_CLIENT_ID": "...",
"SERVICENOW_CLIENT_SECRET": "...",
},
},
},
}
<br>
54 Bundled Skills
Deep domain knowledge for ServiceNow development, organized by category:
| Category | Skills |
|---|---|
| Development | GlideRecord patterns, Script Includes, Business Rules, Client Scripts, UI Builder |
| Integration | REST integration, Integration Hub, Transform Maps, Import/Export |
| Automation | Flow Designer, Approval Workflows, Scheduled Jobs, Change Management |
| Security | ACL patterns, Domain Separation, Instance Security, GRC Compliance |
| ITSM | Incident, Problem, Change, Request, SLA Management |
| Platform | CMDB, Discovery, Performance Analytics, Virtual Agent, Agent Workspace |
| Quality | ATF Testing, Code Review, Widget Coherence, ES5 Compliance |
Extend with MCP & Plugins
<table> <tr> <td width="50%" valign="top">MCP Servers
Connect any MCP-compatible server via stdio, SSE, or streamable HTTP with OAuth support:
{
"mcp": {
"context7": {
"type": "remote",
"url": "https://mcp.context7.com/mcp",
},
"custom-tools": {
"type": "local",
"command": ["node", "./server.js"],
},
},
}
</td>
<td width="50%" valign="top">
Plugins
Extend with npm packages or local plugins that add tools, auth providers, event hooks, and config hooks:
{
"plugin": ["my-snow-flow-plugin", "file://./local-plugin"],
}
</td>
</tr>
</table>
<br>
Headless / Server Mode
Run Snow-Flow as a headless API server for CI/CD pipelines, remote control, or web UI access:
snow-flow serve # Start API server on port 4096
snow-flow serve --port 8080 # Custom port
snow-flow web # Start server + open web interface
snow-flow attach http://host:4096 # Attach TUI to remote server
<br>
Permission System
Fine-grained control over what agents can do, with glob pattern matching and env file protection by default:
{
"permission": {
"bash": "ask",
"write": "allow",
"read": "allow",
"external_directory": "deny",
},
}
<details>
<summary><b>Installation directory priority</b></summary>
<br>
The install script respects the following priority order:
$SNOW_FLOW_INSTALL_DIR— Custom installation directory$XDG_BIN_DIR— XDG Base Directory Specification compliant path$HOME/bin— Standard user binary directory$HOME/.snow-flow/bin— Default fallback
Documentation
For full configuration reference, provider setup guides, and advanced usage:
<br>Contributing
We welcome bug fixes, new provider support, LSP/formatter additions, and documentation improvements. Read our contributing guide before submitting a pull request.
<br>FAQ
<details> <summary><b>What is Snow-Flow?</b></summary> <br>Snow-Flow is an autonomous ServiceNow development agent. It connects AI models to your ServiceNow instance through MCP tools, giving you an intelligent assistant that can develop widgets, query tables, deploy artifacts, manage update sets, and automate tasks — all from your terminal.
</details> <details> <summary><b>How is this different from other coding agents?</b></summary> <br>Snow-Flow is purpose-built for ServiceNow:
- Open source — Elastic License 2.0, fully transparent
- ServiceNow-native — 200+ MCP tools and 54 domain skills designed for ServiceNow
- Provider-agnostic — Works with 20+ AI providers, not locked to one vendor
- Built-in LSP — Language server support for intelligent code assistance
- Terminal-first — A TUI built by terminal enthusiasts with SolidJS + opentui
- Client/server architecture — Remote control, headless mode, and web UI
- Plugin ecosystem — Extend with npm packages or local plugins
Yes. Snow-Flow supports 20+ providers out of the box. Configure your preferred provider in snow-flow.jsonc or through the interactive setup on first launch. You can even switch providers mid-session.
Yes. Snow-Flow connects to any ServiceNow instance via OAuth2 or basic authentication. Configure your instance URL and credentials in the MCP server configuration, and Snow-Flow will have access to all 200+ ServiceNow tools.
</details> <br>Disclaimer
Snow-Flow is an independent, open-source project and is not affiliated with, endorsed by, or sponsored by ServiceNow, Inc. ServiceNow is a registered trademark of ServiceNow, Inc.
This tool requires a valid ServiceNow subscription and uses your own credentials to interact with your ServiceNow instance. Some features (including Flow Designer automation) use undocumented ServiceNow APIs that may change without notice.
Licensed under Elastic License 2.0.
<p align="center"> <b>Join our community</b><br><br> <a href="https://snow-flow.dev"><img alt="Website" src="https://img.shields.io/badge/Website-snow--flow.dev-0ea5e9?style=for-the-badge&logo=safari&logoColor=white" /></a> </p> <p align="center"> If you are building a project that uses "snow-flow" or "snow-code" in its name, please note in your README that it is not built by or affiliated with the Snow-Flow team. </p>