How to download the rConfig Skill for AI agents

The MCP server gives an agent access to rConfig’s tools. A Skill teaches it how to use them well: when to look up a device first, how to compare versions in the right order, how to page through results, and which workflows answer common questions. Install the rConfig Skill alongside your MCP connection and your agent makes fewer wrong turns.

When to use this

Your agent supports Skills (for example Claude) and you want it to handle rConfig requests reliably without long prompts.
You’re setting up several people’s agents and want one package to hand out instead of walking each person through MCP setup.

The Skill doesn’t replace the MCP connection. It contains a ready-made mcp.json that points at your server, and still needs an MCP token.

Prerequisites

Access to the AI Center in rConfig.
An MCP token. Create one on the MCP Server tab before or after downloading. A REST API token won’t work.
An agent that supports Skill packages, or any MCP client if you only want the connection file.

Available Skills

Skill	Status	Tools it covers
rConfig Core	Available	`get_devices`, `get_device`, `get_latest_config`, `get_config`, `get_config_history`, `get_config_diff`

rConfig Compliance will depend on rConfig Core, so install Core first. Until the Compliance Skill ships, your agent can still check compliance through the MCP server’s get_compliance_status tool.

Step-by-step walkthrough

1. Open the Skills tab

Click AI Center in the left sidebar and select the Skills tab.

rConfig V8 Skills tab showing the rConfig Core skill available for download, with compatible agent chips for Claude, Cursor, Cline, and GitHub Copilot

2. Download rConfig Core

Click Download Skill on rConfig Core. Your browser saves rconfig-core-skill-v1.0.1.zip.

3. Look inside the package

Directoryrconfig-core/
- SKILL.md the instructions your agent reads: what rConfig is, when to use the Skill, working rules, and common workflows
- mcp.json connection settings for your server, with a token placeholder
- Directoryreference/
  - connecting.md how to get the endpoint and token, per-client setup, and troubleshooting
  - tools.md inputs and response shapes for each tool
  - workflows.md step-by-step recipes, such as “explain what changed between two versions” and “audit an estate”

mcp.json already contains your server’s MCP endpoint:

{
  "mcpServers": {
    "rconfig": {
      "type": "http",
      "url": "https://<your-rconfig-server>/mcp",
      "headers": { "Authorization": "Bearer <your-mcp-read-token>" }
    }
  }
}

4. Add your MCP token

Replace <your-mcp-read-token> in mcp.json with your MCP token. Keep the edited file out of shared drives and version control, since it now contains a credential.

5. Install it in your agent

How you install depends on the agent:

Claude Code: copy the rconfig-core folder into ~/.claude/skills/ (for you) or .claude/skills/ in a project, and add the MCP server with the command on the MCP Server page.
Claude (claude.ai): upload the zip as a custom Skill. claude.ai runs in the cloud, so it can only reach the MCP server if your rConfig endpoint is reachable from the internet; for internal servers, use Claude Code or Claude Desktop instead.
Cursor, Cline, GitHub Copilot, and other MCP clients: use mcp.json as your client’s MCP configuration, and add SKILL.md to the agent’s rules or instructions.

Check your agent’s own documentation for the current install steps; Skill support changes quickly.

6. Try it

Ask your agent something that needs several steps, such as “Compare the last two configs of core-sw-01 and explain what changed.” With the Skill installed, the agent should look up the device, list its config history, and diff the two newest versions in order, without you explaining how.

How to download the rConfig Skill for AI agents