Using AI Agents

To use ALM for tasks based on AI agents, connect ALM to an AI agent. ALM includes a built-in MCP server for this purpose. The server is installed during software deployment.

To connect the MCP server to an AI agent, open the MCP settings and specify the connection parameters, for example:

{
  "mcpServers": {
    "mcp-alm": {
      "url": "http://localhost/mcp",
      "headers": {
        "Devprom-Auth-Key": "***",
        "Devprom-Base-Url": "http://localhost"
      }
    }
  }
}

We recommend creating a separate ALM account for the MCP server. That way you can control which projects AI agents can access, and set the access level: read, write, delete, and access to individual entities.

While an AI agent is working, it automatically inspects the MCP server capabilities and turns your requests into queries against data stored in ALM. For example, searching for similar texts, searching with the query language, creating a comment, and so on.

Examples of prompts an AI agent can use to call the MCP server:

1.
In the "Development" project, select Issues in the "In progress" state
and for each one create a comment "@user update the task status",
substituting the assignee (Owner field) for user.

2.
In the "Development" project, find Requirements that are similar in meaning to
"eat some more of these soft French rolls".


Search results