VS Code and Copilot

This page is about Copilot's agent mode in VS Code calling the coordigent.* MCP tools. The Coordigent extension for VS Code, which shows your agents in the editor, is separate: see Extension manager.

coord login and coord connect do not register VS Code on their own. You name it, either once for your VS Code profile or in a repo.

Set it up

For every repo, once per machine:

$ coord install vscode --user

Or, to write the files into one repo for a team that commits them, run this inside it:

$ coord connect --repo-files --clients vscode

Without --clients, coord connect --repo-files writes VS Code's files when the repo already has a .vscode/mcp.json or a .github/copilot-instructions.md, or when it finds VS Code on this machine (~/.vscode, or code or code-insiders on your PATH) and you say yes to its one question. A .vscode/ folder on its own does not count, since most repos have one for editor settings.

What gets written

In the repo

.vscode/mcp.json, under VS Code's servers key rather than mcpServers:

{
  "servers": {
    "coordigent": {
      "type": "stdio",
      "command": "sh",
      "args": ["-c", "exec \"${TOWER_BIN_DIR:-$HOME/.tower/bin}/coord-mcp\" \"$@\"", "coord-mcp"],
      "env": { "COORDIGENT_REPO": "your-org/your-repo" },
      "cwd": "${workspaceFolder}"
    }
  }
}

The server is found through $HOME, so the file is identical on every teammate's machine. The merge keeps your comments and any servers already listed. No token is stored here: coord-mcp reads your login from ~/.tower/config.json.

.github/copilot-instructions.md gets the coordination rules, between  and  markers. The rest of the file is yours.

coord connect --repo-files also writes Claude Code's repo files; see Claude Code.

In your VS Code profile

coord install vscode writes the repo files above without the rest of coord connect. With --user it writes your profile's mcp.json instead, without the cwd line:

System	File
macOS	`~/Library/Application Support/Code/User/mcp.json`
Linux	`~/.config/Code/User/mcp.json`
Windows	`AppData/Roaming/Code/User/mcp.json` under your home folder

Pass --config-path <file> if you use a VS Code profile other than the default. At user scope the rules are printed for you to add through Chat: New Instructions File in your user profile. Both forms also print an equivalent code --add-mcp command, if you would rather have VS Code write its own profile entry.

$ coord install vscode --user

Update

$ coord update

This downloads the current coord and coord-mcp and restarts the daemon. Restart VS Code, or restart the server from MCP: List Servers, to load the new coord-mcp.

Check it

$ coord doctor vscode

coord doctor lists vscode with a ✓ and the config file it found. A ✗ line names the command that fixes it. In VS Code, run MCP: List Servers, pick coordigent, choose Start, then ask Copilot to call coordigent.status.

What works and what degrades

VS Code supports MCP resources and prompts. Sampling, elicitation, roots, server notifications and resource subscriptions stay off. Coordigent does not need them.

Files in use. Coordigent has no hook in Copilot, so it does not see what a Copilot agent edits. The agent still appears in the project once it calls a coordigent.* tool, and can read status and send messages.
Messages. A message or conflict waits until the agent's next coordigent.* tool call, and arrives as a ⚠ COORD: banner on its result. coordigent.message.inbox reads the rest.

Remove

To remove Coordigent from this machine entirely, run coord uninstall (add --dry-run to see the plan first). It keeps anything you edited by hand. It cleans each repo's .vscode/mcp.json, but not your profile's mcp.json: delete that entry yourself if you used --user.

To remove only VS Code's setup by hand, delete the coordigent entry from servers in .vscode/mcp.json or your profile's mcp.json, and the marked block in .github/copilot-instructions.md.

Extension manager covers the Coordigent extension for VS Code.
MCP tools lists every coordigent.* tool the agent can call.

#Set it up

#What gets written

#In the repo

#In your VS Code profile

#Update

#Check it

#What works and what degrades

#Remove

#Next

Set it up

What gets written

In the repo

In your VS Code profile

Update

Check it

What works and what degrades

Remove

Next