msd-core/docs/how-to/connect-gsd-mcp-server.md

How to connect a host to the GSD companion MCP server

This guide shows you how to make a MCP-capable host (Claude Code, Codex, OpenCode, VS Code, Antigravity CLI, Cursor, Cline, Hermes, Augment Code) drive GSD — run GSD commands and read/write .planning/ state — through the companion MCP server, with no bespoke plugin.

Once connected, three tools appear in the host alongside its others: gsd_invoke_command, gsd_read_state, gsd_write_state. The server also serves a read-only catalog of GSD's own content — workflows and references as MCP resources, and the /gsd-* commands as MCP prompts — so a host can browse and pull that content directly instead of shelling out to the CLI. (For the tool contracts, see the reference section below; for why this server exists and its trust model, see ADR-1239 and the capability trust model.)

1. Add the server to your host's MCP config

The entry shape is the same everywhere; only the config file and key differ by host.

{
  "gsd": {
    "command": "npx",
    "args": ["-y", "@opengsd/gsd-core", "gsd-mcp-server"],
    "cwd": "/abs/path/to/your/project"
  }
}

• Claude Code / Codex / Cursor / Cline / Hermes — under the host's mcpServers object (project or user config).
• Augment Code — under the mcpServers block of its own settings.json (not a standalone MCP config file, unlike Antigravity) — global at ~/.augment/settings.json, project-local at .augment/settings.json. GSD's installer configures this entry automatically (--augment installs).
• VS Code — in the workspace MCP servers list.
• Antigravity — under the mcpServers block of its standalone mcp_config.json profile (not embedded in settings.json) — global at ~/.gemini/antigravity/mcp_config.json (or the sibling antigravity-ide/antigravity-cli dir GSD resolved into), project-local at .agents/mcp_config.json. GSD's installer configures this entry automatically (--antigravity installs).
• OpenCode — under the mcp key (not mcpServers), in ~/.config/opencode/opencode.jsonc (global) or ./opencode.json (project). The entry shape also differs — see below.
• Kilo Code — an OpenCode fork; also under the mcp key (not mcpServers), in ~/.config/kilo/opencode.jsonc (global) or ./opencode.json (project). Same entry shape as OpenCode.

Set cwd to the project whose .planning/ you want GSD to manage — the server resolves state paths against it.

OpenCode / Kilo entry shape

OpenCode (and Kilo, which shares OpenCode's config schema) use a type/command/timeout entry under the mcp key instead of the generic command/args/cwd form above:

{
  "mcp": {
    "gsd": {
      "type": "local",
      "command": ["npx", "-y", "@opengsd/gsd-core", "gsd-mcp-server"],
      "timeout": 10000
    }
  }
}

2. Restart the host

On startup the host performs the MCP initialize handshake. The response advertises tools, resources, and prompts capabilities, so the three GSD tools become callable and the host can also list the served catalog (resources and prompts) described below. The server never advertises resources.subscribe or listChanged — the catalog is fixed for the life of the server process, so there is nothing to subscribe to.

3. Verify

Ask the host to read an existing planning file:

{ "name": "gsd_read_state", "arguments": { "path": "/abs/path/to/your/project/.planning/STATE.md" } }

It returns the file's contents. gsd_invoke_command takes {family, subcommand, args} and returns the command-routing hub's structured result (the same shape gsd-tools produces).

4. Browse the catalog (resources and prompts)

The server also exposes GSD's own content tree as MCP resources and the commands/gsd/*.md command set as MCP prompts. This is additive: the file-copy install (the default for every runtime) is unchanged, and the catalog only adds a way for an MCP-capable host to read the same content directly over the protocol.

List and read a resource

List available resources (paginated — ask the host to follow nextCursor until it is absent):

{ "name": "resources/list", "arguments": { "cursor": null } }

Each entry has a gsd://<segment>/<relpath> URI, where <segment> is workflows, references, or commands, and <relpath> is the file's path within that segment (for example, gsd://workflows/plan-phase.md, gsd://references/untrusted-input-boundary.md, or gsd://commands/plan-phase.md). Commands appear in both surfaces: read one as a resource to get its raw markdown, or get it as a prompt to have the host treat it as an invocable message. Read one by URI:

{ "name": "resources/read", "arguments": { "uri": "gsd://workflows/plan-phase.md" } }

An unknown or unrecognized URI (including any path-traversal or absolute-path attempt) returns a JSON-RPC error rather than an empty or partial result.

List and get a prompt

{ "name": "prompts/list", "arguments": {} }

Each entry is keyed by its bare command name — plan-phase, not a path. Get one:

{ "name": "prompts/get", "arguments": { "name": "plan-phase" } }

An unknown prompt name returns a JSON-RPC error. prompts/get accepts an arguments object but ignores it — no shipped command template takes injected arguments today.

Composed vs. verbatim content

Workflow resources (gsd://workflows/…) are served composed — with <!-- gsd:section --> markers stripped — exactly as the installed file tree gets them, while reference and command content is served verbatim, because some reference and command docs use that marker syntax as a documented example rather than a real marker.

If something does not work

• command not found: gsd-mcp-server — invoke via npx as shown above, or install the package globally first (npm i -g @opengsd/gsd-core).
• gsd_read_state fails with ENOENT — the path is resolved literally; pass an absolute path under the project's .planning/.
• The host lists no GSD tools — confirm the server starts in isolation: npx @opengsd/gsd-core gsd-mcp-server then send an initialize request on stdin; it writes a protocolVersion response and exits on EOF.
• You manage multiple projects — register one gsd entry per project with a distinct name and cwd; the server is stateless across projects.

Reference — the three tools

Tool	Arguments	Returns
gsd_invoke_command	{family: string, subcommand: string, args?: unknown[]}	the command-routing hub result ({ok, …}) as JSON text
gsd_read_state	{path: string}	the file contents as text
gsd_write_state	{path: string, content: string}	{ok: true, path} as JSON text

Errors from a tool are returned as MCP tool errors (isError: true), not as JSON-RPC protocol errors — the host surfaces them in its normal tool-failure UX.

Reference — the served catalog

Method	Arguments	Returns
resources/list	{cursor?: string}	{resources: [{uri, name, title, description, mimeType}], nextCursor?: string}
resources/read	{uri: string}	{contents: [{uri, mimeType, text}]}
prompts/list	{}	{prompts: [{name, title, description}]}
prompts/get	{name: string, arguments?: object}	{description, messages: [{role: "user", content: {type: "text", text}}]}

Errors from the catalog (unknown URI, unknown prompt name, malformed cursor, a refused path-traversal attempt) are returned as JSON-RPC protocol errors, not MCP tool errors — unlike the three tools above.