Allow to customize MCP server configurations

[9inelemons]

Currently I am running multiple containers inside docker and to run MCP server I need to customize settings for my agent but want to have some feature like "building" guidelines want to have option to customize MCP server settings so its available by any agent like its done for guidelines.

Obviously I can just do it and commit and store in repo but my team can use different agents so I want them to have option just to use it and dont implement MCP server config by themselves

Basically my mcp server config looks like this

{
  "mcpServers": {
    "laravel-boost": {
      "command": "docker",
      "args": [
        "compose",
        "-f",
        "docker-compose.yml",
        "exec",
        "-T",
        "container.name",
        "php",
        "artisan",
        "boost:mcp"
      ],
      "env": {
        "FOO": "BAR",
        "BAZ": "FISH"
      }
    }
  }
}

And also I need to provide custom cwd because docker compose file is located in different directory

[pushpak1300]

set env BOOST_PHP_EXECUTABLE="docker compose -f docker-compose.yml exec -T container.name php" on latest version and run boost:install.

[9inelemons]

set env BOOST_PHP_EXECUTABLE="docker compose -f docker-compose.yml exec -T container.name php" on latest version and run boost:install.

@pushpak1300 First of all, ty for answer! Secondly, you are completely wrong with your comment in some reasons:

1. You mistyped BOOST_PHP_EXECUTABLE, its actually called BOOST_PHP_EXECUTABLE_PATH.
2. If you will try to set env variable like this:

BOOST_PHP_EXECUTABLE_PATH: "FOO=BAR docker compose -f docker-compose.yml exec -T container.name php"

It will produce garbage:

{
    "mcpServers": {
        "laravel-boost": {
            "command": "FOO=BAR",
            "args": [
                "docker",
                "compose",
                "-f",
                "docker-compose.yml",
                "exec",
                "-T",
                "container.name",
                "php",
                "artisan",
                "boost:mcp"
            ]
        }
    }
}

3. Even without setting variables anyway you have to suffer with CWD (current working directory). For codex its acually will generate current working directory inside docker container which may not work in some cases.

• For example if you are using MacOS and your container's working directory is /app MCP server for codex will try to directory /app which is not a thing on MacOS and cant be created

---

Summarizing I want these features:

1. Customize CWD for configuring MCP servers
2. Ability to set environment variables
3. Ability to customize command and args for MCP server in more convenient way

[github-actions]

Thank you for reporting this issue!

As Laravel is an open source project, we rely on the community to help us diagnose and fix issues as it is not possible to research and fix every issue reported to us via GitHub.

If possible, please make a pull request fixing the issue you have described, along with corresponding tests. All pull requests are promptly reviewed by the Laravel team.

Thank you!

[jonaaix]

I have a slightly different approach:
The safest way is to create a bridge script mcp-boost.sh with this content. This allows the repo to work with any platform or docker installation path.

#!/bin/bash

# Bridge for Laravel Boost
cd "$(dirname "$0")"
$(which docker) compose exec -T php php artisan boost:mcp

Then set the MCP config like this:

{
   "mcpServers": {
      "laravel-boost": {
         "command": "./mcp-boost.sh",
         "args": []
      }
   }
}

However it is incredibly annoying that the customized command is overwritten every time I run boost:update. There should be a centralized config parameter in the ./boost.json to define the command and arg properties.

[pushpak1300]

However it is incredibly annoying that the customised command is overwritten every time I run boost:update

I think this should not happen on Boost 2.0 we disabled this behaviour long time ago. Can you confirm if you're on latest version and update is updating the mcp config ?

[jonaaix]

@pushpak1300 Sorry I was wrong. Update does not overwrite it. I had ran install again to setup the Filament Blueprint integration.

[pushpak1300]

Thank you for raising this.

This appears to be a very specific use case that is likely to affect only a small subset of Boost users. In this situation, manually updating the configuration once should be sufficient. Running php artisan boost:update will not modify your existing setup, so you can safely apply the update and continue as normal.

Since Boost already exposes the configuration file, it can be edited directly or committed to git to avoid repeating the same changes in the future. Given how limited this use case is, adding built-in support would add unnecessary complexity for something that can already be solved by editing a simple file.

For now, I’ll hold off on implementing this. If more users run into the same issue, we can revisit it later.