Splynx Add-ons MCP Server

The Splynx Add-ons MCP Server packages Splynx add-on development knowledge - architecture, hooks, API reference, and code patterns - directly into an AI coding assistant’s context. Connect it to Claude Code and the assistant stops guessing at Splynx conventions: it can generate an InstallController, wire up a hook handler, or scaffold a whole add-on skeleton that already matches how Splynx add-ons are structured.

Why use it

• Skip re-reading the add-on architecture guide for every new project - the directory structure, lifecycle, and conventions are already loaded into the assistant’s context.
• Scaffold a working skeleton in one request - files, InstallController, hooks, config.json - instead of copying an existing add-on and stripping it down by hand.
• Get code that matches Splynx’s own conventions - generated controllers, models, and views follow the same Yii2 patterns used across Splynx’s built-in add-ons.
• Guided workflows for the integration types ISPs build most often - payment gateways, accounting sync, e-invoicing - that walk through the parts that are easy to get wrong, like OAuth2 flows, entity pairing, or hash-chaining a submission for a tax authority.

What you can ask it to do

The server bundles three things: reference documentation, code generators, and guided workflows. Nothing here is fetched live - it all ships with the server.

Documentation, embedded (19 topics)

Grouped here for readability - the server ships all 19 as a flat reference set.

Code generators (9 tools)

Guided workflows (9 prompts)

Three ways to run it, depending on how your team works.

Option 1 - Local (stdio)

Build it once, then point Claude Code at the compiled server:

git clone https://github.com/ruslan-splynx/splynx-addons-mcp-server
cd splynx-addons-mcp-server
npm install
npm run build

Add it to your Claude Code MCP configuration - either a project-level .mcp.json (shared with your team) or your user-level ~/.claude.json:

{
  "mcpServers": {
    "splynx-addons": {
      "command": "node",
      "args": ["/path/to/splynx-addons-mcp-server/dist/index.js"]
    }
  }
}

Option 2 - Self-hosted, remote HTTP

Run it as a server your whole team can point to:

node dist/index.js --http --port=3000 --host=0.0.0.0

{
  "mcpServers": {
    "splynx-addons": {
      "type": "url",
      "url": "https://your-host.example.com/mcp"
    }
  }
}

Option 3 - Docker

docker build -t splynx-addons-mcp .
docker run -d --name splynx-mcp --restart always -p 3000:3000 splynx-addons-mcp

A single request walks through the full build. For example:

• No live customer data. The server works with your own codebase and a bundled copy of Splynx’s add-on documentation - nothing is fetched from a running Splynx instance.
• No admin API calls. It never reads or writes anything through your Splynx admin API - for that, see the Splynx MCP Server.
• Review before you deploy. Treat generated code like any AI-assisted output - review it before installing it on a production Splynx instance.

The server doesn’t replace Splynx’s existing developer references - it packages them for your AI assistant so you don’t have to keep them open in another tab. For anything not covered here, these remain the source of truth:

• Addon Howto - the official add-on development guide
• Addon Skeleton - the starter template
• Splynx API v2.0 reference
• Splynx Knowledge Base