MCP servers

Give Nebula your own tools by connecting Model Context Protocol servers to the CLI, at user or project scope, with approval for shared configs.

On this page

The CLI can use tools from any Model Context Protocol server: local commands over stdio, or remote servers over HTTP or SSE.

Add a server

Local stdio server

breachline mcp add asset-lookup ./bin/asset-lookup-mcp

Remote HTTP server with a header

breachline mcp add internal-assets https://mcp.example.com/mcp \
  --transport http \
  --header "Authorization: Bearer $ASSETS_TOKEN"

Option	Effect
`--transport`	`stdio`, `http` or `sse`.
`--scope`	`user` (default config) or `project` (`.mcp.json`).
`--env KEY=VALUE`	Environment variable for a stdio server. Repeatable.
`--header "Key: Value"`	HTTP header for a remote server. Repeatable.

Run breachline mcp add with no arguments for an interactive wizard, or use breachline mcp add-json <name> <json> to paste a definition.

Scope	Stored in	Trust
`user`	Your user configuration	Written by you; connects automatically.
`project`	`.mcp.json` in the repository	Shared with everyone who clones it, so it starts pending and does not connect until you run `breachline mcp approve <name>`.

Command	Effect
`breachline mcp list`	Servers and their status (`--json` for scripts).
`breachline mcp get <name>`	One server’s config, status and capabilities.
`breachline mcp auth [name]`	Sign in to an OAuth-enabled server.
`breachline mcp remove <name>`	Remove and disconnect.
`breachline mcp edit`	Edit the config file directly.

MCP servers you connect in the console are covered in MCP servers.