
What is MCP in Antigravity?
MCP (Model Context Protocol) in Antigravity is how you connect an Google Antigravity agent to external tools like databases, GitHub, Figma, or custom APIs, so the agent can read real data and take action instead of waiting for you to paste information into the chat. Along with MCP come skills (directories containing SKILL.md that guide an agent to do one thing) and plugins (packages that bundle skills, MCPs, sub-agents, rules, and hooks). This post makes no promise that any utility is safe or effective; we have not reviewed the code of any plugin in the Marketplace, and only describe what the official documentation and agy --help version 1.4.12 say, read on 03/11/2026. The post covers how to add MCP servers, skills, plugins, hooks, risks, and comparison with Claude Code and Codex. A general introduction is in the post What is Antigravity?.

Three layers to extend an agent in Antigravity: MCP, skill, and plugin, sitting below the permissions layer. Diagram: OPCAIOS.
How are MCP, skill, and plugin different in Antigravity?
MCP, skill, and plugin are three different layers to extend an Antigravity agent, according to the pages /docs/mcp, /docs/skills, and /docs/plugins (read on 03/11/2026). MCP adds tools and data to an agent: an MCP server is a program running locally (stdio) or a remote service that exposes tools for the agent to call. Skill adds methods to an agent: a directory with SKILL.md that holds a process, checklists, and may include scripts; the agent only sees the name and description until it needs to read the full text. Plugin is a way to package and distribute: a directory with plugin.json that bundles skills, MCP servers, sub-agents, rules, and hooks, installed with one command and toggled on and off together. If you are new, see What is MCP?.
| Layer | Adds to agent | Main file |
|---|---|---|
| MCP server | Tools, external data | mcp_config.json |
| Skill | Process, knowledge | SKILL.md |
| Plugin | Package combining multiple parts | plugin.json |
How do I add an MCP server in the Antigravity app?
To add an MCP server in the Antigravity app, go through the built-in MCP Store, according to the page /docs/mcp (read on 03/11/2026):
- Press Settings in the bottom left corner.
- Select Customizations, see the Installed MCP Servers section.
- Press Add MCP to open the MCP Store.
- Find the server, press Add, log in if asked.
The Store is organized by category like databases (Supabase, Neon, BigQuery), developer tools (GitHub, Linear, Vercel), and design (Figma, Canva). Servers not in the Store can be added by editing mcp_config.json directly; the Antigravity IDE has a similar Store.
How do I add an MCP server with agy mcp add?
To add an MCP server with Antigravity CLI, there are two ways: open the manager with /mcp in the interactive interface, or use agy mcp from the shell, according to agy mcp add --help version 1.4.12 (read on 03/11/2026). The syntax is server name, then the run command (stdio server) or URL (http server), then parameters. The --env (-e) flag passes environment variables, --header (-H) adds HTTP headers, --type (-t) chooses stdio or http; http and https URLs are recognized automatically so you usually don't need --type. Note: flags must come before the server name, and use -- before the command if the command starts with a dash. Other subcommands are list, remove, enable, disable. According to the page /docs/mcp, the CLI reads servers from a global file for all projects and from .agents/mcp_config.json in the workspace; active plugins can also add servers. Example taken from the help text:
agy mcp add fs npx -y @modelcontextprotocol/server-filesystem /work
agy mcp list
What should I pay attention to when configuring mcp_config.json?
Antigravity's mcp_config.json file uses the same structure across the application, IDE, CLI and plugins: the root is an mcpServers object, each key is a server name, according to the /docs/mcp page (read on 03/11/2026). Each server has exactly one connection type: command for stdio servers or serverUrl for remote servers. Remote servers must use Streamable HTTP; the old HTTP+SSE type and httpUrl key are no longer supported. Some useful optional keys:
enabledTools,disabledTools: let the agent see only certain tools, reducing risk and context size.disabled: temporarily turn off a server without deleting its configuration.timeoutSeconds: default wait for connection is 5 minutes, each tool call is 3 minutes; negative values disable limits.headers,oauth,authProviderType: authenticate via headers, OAuth, or Google Cloud credentials.
As for permissions, MCP tools don't have a default configuration in ask mode. You can allow them at three levels: mcp(server/tool), mcp(server/*) and mcp(*). The safe approach is to only allow read-only tools, keeping ask mode for write or delete tools. API keys in headers are stored in lowercase, so don't put this file in a public code repository. See more at safety mode in Antigravity.
How do Skills work in Antigravity?
A Skill in Antigravity is a directory containing a required SKILL.md file, optionally accompanied by scripts/, examples/, resources/, according to the /docs/skills page (read on 03/11/2026). The YAML frontmatter has required description and optional name. Skills load gradually: each turn, the agent only sees the name and description of all skills; when a skill matches the request, the agent reads the entire SKILL.md and follows it. So descriptions need clear keywords. Workspace-level skills are placed in .agents/skills/, global skills are shared across the application and CLI. In the CLI, each skill becomes its own slash command, for example the deploy-staging skill becomes the /deploy-staging command; to add a skill while the CLI is running, type /skills reload. According to the /docs/migration/workflows-to-skills page, old workflows stopped working from 19/10/2026 and the /migrate-workflows command helps migrate to skills; Antigravity follows the Agent Skills standard from May 2026, so conforming skills can be reused in other tools.
What are Plugins and the Plugin Marketplace?
A Plugin in Antigravity is a versioned package that combines five types of components: skills, MCP servers, custom agents and sub-agents, rules, hooks, according to the /docs/plugins page (read on 03/11/2026). The Plugin Marketplace is Google's and partners' plugin catalog, announced on the Antigravity blog on 29/10/2026, available in the application and CLI. In the app, open Customizations then Marketplace; in the CLI, type /plugin. Plugins installed in the app automatically sync to the CLI. From the shell, according to agy plugin --help:
| Command | Action |
|---|---|
agy plugin install <name> |
Install from the official Marketplace, GitHub URL, or local directory |
agy plugin import gemini or claude |
Import Gemini CLI extension or Claude Code plugin |
agy plugin list --json |
List plugins, components, paths |
agy plugin validate --strict |
Validate manifest and files |
A plugin's MCP tools are named as <plugin>_<server>. Plugins requiring API keys are installed in disabled state until you enter the keys.
What are Hooks in Antigravity used for?
Hooks in Antigravity are shell commands that run at fixed points in the agent loop, declared in hooks.json at workspace, global, or plugin level, according to the /docs/hooks page (read on 03/11/2026). There are six events: SessionStart, PreToolUse, PostToolUse, PreInvocation, PostInvocation and Stop. The PreToolUse hook returns a decision field like allow, deny (block immediately), ask, force_ask, and can modify parameters before the tool runs. This is where you set up safety nets for MCP; for example, a hook could block all delete commands in a data directory. In the CLI, /hooks lists active hooks.
What are the risks when installing MCP, unfamiliar skills, and plugins?
The biggest risk of unfamiliar utilities is that they run with the permissions of the user on your machine. This is an analysis from OPCAIOS based on how the documentation describes the mechanism, not a warning extracted from Google:
- Hooks in plugins run automatically when the plugin is enabled, meaning a plugin can execute shell commands without the agent calling it.
- MCP server stdio is a local program: installing from an unfamiliar package is like installing unfamiliar software.
- SKILL.md is instructions for the agent: a malicious skill can instruct the agent to send data outside.
- Data through MCP can contain embedded commands targeting the agent (command injection).
Ways to reduce risk: prioritize official plugins, read hooks.json and SKILL.md before installing from GitHub, limit tools using enabledTools, and use narrow-scoped keys. See security when using AI to write code.
How does Antigravity compare with Claude Code and Codex?
All three tools have MCP and plugin management commands, differing in syntax and storage location, according to the CLI help on our machine (Antigravity CLI 1.4.12, Claude Code 2.1.317, Codex CLI 0.162.1, dated 03/11/2026):
| Task | Antigravity CLI | Claude Code | Codex CLI |
|---|---|---|---|
| Add MCP | agy mcp add |
claude mcp add |
codex mcp add |
| Config file | mcp_config.json |
Claude Code's own config | config.toml |
| Plugin | agy plugin |
claude plugin |
codex plugin (has marketplace management) |
| Import from other tools | agy plugin import gemini/claude |
claude mcp add-from-claude-desktop |
Not seen in help |
Details on the Codex side are in the post MCP, skills and hooks in Codex.
OPCAIOS's experience
The Antigravity application on OPCAIOS's main Mac declares 14 MCP servers; CLI agy 1.4.12 has one additional browser skill installed, and the user-level skill folder has one orchestration skill for the terminal management tool we use. Hooks exist at points before and after the agent and before and after the tool, but only for reporting session status; we don't yet have hooks that block commands. There is exactly one global workflow file but it's an empty template, not actually used. The command agy plugin list on 03/11/2026 shows the CLI has 6 plugins, all from Google's official source (for example, Chrome DevTools plugin, Firebase); we haven't confirmed which plugins come pre-installed, which ones the agent installed, haven't installed any third-party plugins, haven't run agy plugin import, and haven't measured the effectiveness of each MCP server.
Frequently Asked Questions
Are MCP Store and Plugin Marketplace the same?
Not entirely. MCP Store in the app and IDE is a list of MCP servers to install individually. Plugin Marketplace, announced on 29/10/2026, is a catalog of plugins, where each plugin can combine skills, MCP servers, sub-agents, rules and hooks. A plugin can contain an MCP server, but a standalone MCP server is not a plugin.
Can I use old SSE-style MCP servers?
According to the /docs/mcp page read on 03/11/2026, remote MCP servers must use Streamable HTTP; the old HTTP+SSE style and the httpUrl key are no longer supported. If a server only has SSE, you need to ask the provider for a Streamable HTTP version or use a local bridge in stdio form. The correct URL key is serverUrl, though the url key is also accepted.
Can skills written for Claude Code be used in Antigravity?
Antigravity follows the open Agent Skills standard, with the same directory format containing SKILL.md with description frontmatter, so in principle a properly standardized skill can be reused. You need to place it in .agents/skills/ or a global skills folder, and check whether the accompanying scripts have dependencies on Claude Code-specific tools. Claude Code plugins can be imported with agy plugin import claude; we haven't tested this yet.
How do I know if a plugin is safe?
There's no foolproof way. At minimum you should read plugin.json, hooks.json, the SKILL.md files and the MCP server code before installing, especially for plugins from GitHub. agy plugin validate --strict only checks structure, not intent. After installation, use agy plugin list --json to see what components the plugin adds.
Where should I start with MCP and skills in Antigravity?
Start with a read-only MCP server from MCP Store, for example reading a database schema as a test, then write a small skill for a repeating weekly task. If you want to learn how to connect agents with external tools under control, see OPCAIOS's AI application training course.
Source: Google Antigravity documentation at antigravity.google/docs (mcp, skills, plugins, marketplace, build-plugins, hooks, migration/workflows-to-skills), blog "Plugin Marketplace in Antigravity" (29/10/2026), help for agy mcp, agy plugin version 1.4.12, claude mcp, claude plugin, codex mcp, codex plugin (read on 03/11/2026); not verified: plugin safety in Marketplace, actual results of agy plugin import. Experience is from OPCAIOS.
Translated from Vietnamese with AI. Read the original.
Want to apply this in your business?
Book a free 30-minute call (Vietnam time, in English or Vietnamese).
