4.11.1 Examples of collaboration with AI agents

This section provides examples of connection settings for the MCP server and specific prompts.

NotesNotes

This information is current as of February 2026. The procedures here are subject to change due to updates to coding agent specifications or other factors. See the latest information as needed before proceeding.


(1) Example of connection settings for the Ops I MCP server

Change the endpoint (URL) used according to your purpose. For the endpoint, see “API reference overview > MCP server” in “JP1 Cloud Service/Operations Integration API Reference”.

If you want to automatically generate or modify code, select a coding agent.
Ops I has verified operation with the following coding agents:

  • GitHub Copilot agent mode (Claude Sonnet 4.6)
  • Claude Code (Claude Sonnet 4.6)
TipsTips
Using high-precision AI models released in March 2026 or later might provide more accurate results.

Examples of connection settings for each coding agent are shown below.

[When using Visual Studio Code and GitHub Copilot]

  1. Create a file named ".vscode/mcp.json" in your workspace, enter the following text, and save it.

    {
      "servers": {
        "opsi-manual": {
          "url": "Ops I URL/ai/v1/manual/mcp",
          "type": "http",
          "headers": {
            "X-OpsI-Token": "Ops I token"
          }
        }
      }
    }


  2. A [Start] button will appear above "opsi-manual" in the mcp.json file you created. Click it.

[When using Claude Code]

Run the following command:

claude mcp add --transport http opsi-manual --scope user Ops I URL/ai/v1/manual/mcp --header "X-OpsI-Token: Ops I token"



(2) Examples of agent skill installation

The procedure is the same whether you use Visual Studio Code and GitHub Copilot or use Claude Code.
The steps are outlined below:

  1. Download the "jp1cs-opsi-customization-guide" folder.
    For the storage location, see "(Table) Agent skill storage location".
  2. Create the "~/.claude/skills" directory.
  3. Place the folder downloaded in Step 1 into the directory created in Step 2.

(3) Example prompts

(a) Ticket and knowledge article operations

[When creating a ticket]

Multiple users have reported not being able to log in to the cloud service.
When users attempt to log in to the cloud management window, the message "Authentication failed" appears, and they are not able to log in.
Reports indicate that although the password should be correct, users are not able to log in no matter how many times they try.

Create an incident ticket with the details above in the body, a summary of the details in the title, and priority/impact/severity set to High/High/High.



(b) AI-assisted coding of Ops I definition files

[When creating a workflow]

■ Prompt

The purpose of this project is to implement the Ops I workflow.
The workflow performs the VM rental process.
Load the workflow specification document ("workflow_specification.md") and proceed with the implementation.

■ Specification document
Place the specification document in a location accessible to the coding agent. Also, save it in a format supported by the coding agent you are using, such as text or images. Depending on the file format of the specification document, additional agent skills might be required to load it.

NotesNotes

Entering this prompt once does not guarantee that development will be completed.
If incorrect output is generated, you will need to provide additional prompts to the AI and gradually revise and improve the code.