<p align="center">
<a href="https://sanity.io">
<img src="https://cdn.sanity.io/images/3do82whm/next/d6cf401d52c33b7a5a354a14ab7de94dea2f0c02-192x192.svg" />
</a>
<h1 align="center">Sanity Agent Toolkit</h1>
</p>
Collection of resources to help AI agents build better with [Sanity](https://www.sanity.io). Supports Cursor, Claude Code, Codex, VS Code, Lovable, v0, Replit, OpenCode, and any other editor/agent compatible with MCP or [Agent Skills](https://agentskills.io).
---
## Features
- **MCP server:** Direct access to your Sanity projects (content, datasets, releases, schemas) and agent rules.
- **Agent skills:** Comprehensive best practices skills for Sanity development, content modeling, SEO/AEO, and experimentation. Includes 25 integration/topic guides and 26 focused best-practice rules.
- **Agent Plugin:** `plugin.json`, `skills/`, and `mcp.json` components for any Agent Plugins-compatible client.
- **Claude Code plugin:** MCP server, agent skills, and slash commands for Claude Code users. Available on the [official Anthropic plugin marketplace](https://claude.com/plugins/sanity).
- **Cursor plugin:** MCP server, agent skills, and commands on the [Cursor Marketplace](https://cursor.com/marketplace/sanity).
- **Codex plugin:** MCP server and agent skills for [OpenAI Codex](https://developers.openai.com/codex) users.
---
## Get started
Choose your path based on how you want agents to work with Sanity:
1. **MCP server** — Give your agent always up-to-date rules and full access to your Sanity projects. No local files to maintain. Works with Cursor, VS Code, Claude Code, Lovable, v0, Replit, OpenCode, and other MCP-compatible clients.
2. **Agent skills** — Install best practices skills for Sanity, content modeling, SEO/AEO, and experimentation. Works with Cursor, Claude Code, and any [Agent Skills](https://agentskills.io)-compatible agent.
3. **Plugin** — Install the Sanity plugin for Cursor, Claude Code, or any Agent Plugins-compatible client. Bundles MCP server, agent skills, and commands.
4. **Manual installation** — Copy the skill references locally for offline use. You'll need to update them yourself.
### Option 1: Install MCP server (recommended)
Give agents direct access to Sanity projects and always up-to-date agent rules via the MCP server.
#### Quick install via Sanity CLI
Run in terminal to detect and configure MCP for Cursor, Claude Code and VS Code automatically:
```bash
npx sanity@latest mcp configure
```
Uses your logged-in CLI user for authentication — no manual tokens or OAuth needed.
#### Client-specific instructions
<details>
<summary><strong>Cursor</strong></summary>
One-click install:<br>
[](https://cursor.com/en-US/install-mcp?name=Sanity&config=eyJ0eXBlIjoiaHR0cCIsInVybCI6Imh0dHBzOi8vbWNwLnNhbml0eS5pbyJ9)
Or manually: Open **Command Palette** (`Cmd+Shift+P` / `Ctrl+Shift+P`) → **View: Open MCP Settings** → **+ New MCP Server** → add to `mcp.json`:
```json
{
"mcpServers": {
"Sanity": {
"type": "http",
"url": "https://mcp.sanity.io"
}
}
}
```
</details>
<details>
<summary><strong>Claude Code</strong></summary>
Run in terminal. Authenticate with OAuth on next launch:
```bash
claude mcp add Sanity -t http https://mcp.sanity.io --scope user
```
</details>
<details>
<summary><strong>Codex</strong></summary>
Run in terminal. Authenticate with OAuth on next launch:
```bash
codex mcp add Sanity --url https://mcp.sanity.io
```
Or manually add to `~/.codex/config.toml`:
```toml
[mcp_servers.Sanity]
url = "https://mcp.sanity.io"
```
</details>
<details>
<summary><strong>VS Code</strong></summary>
Open **Command Palette** (`Cmd+Shift+P` / `Ctrl+Shift+P`) → **MCP: Open User Configuration** → add:
```json
{
"servers": {
"Sanity": {
"type": "http",
"url": "https://mcp.sanity.io"
}
}
}
```
</details>
<details>
<summary><strong>Lovable</strong></summary>
Sanity is available as a prebuilt chat connector in Lovable:
1. Open **Connectors** → **Chat connectors**
2. Select **Sanity**
3. Click **Connect** and sign in to authorize your Sanity account
In your next prompt, reference your Sanity project or ask the agent to read your schema.
See the [Lovable MCP documentation](https://docs.lovable.dev/integrations/mcp-servers) or [Sanity + Lovable guide](https://lovable.dev/connect/sanity) for more details.
</details>
<details>
<summary><strong>v0</strong></summary>
In the prompt input field, click **Prompt Tools** → **MCPs** → **Add New** → Select **Sanity** → **Authorize** → Authenticate with OAuth.
</details>
<details>
<summary><strong>Replit</strong></summary>
Go to [Integrations Page](https://replit.com/integrations) → scroll to **MCP Servers for Replit Agent** → **Add MCP server** → Enter `Sanity` as name and `https://mcp.sanity.io` as Server URL → **Test & Save** → Authenticate with OAuth.
</details>
<details>
<summary><strong>OpenCode</strong></summary>
Add to your `opencode.json`:
```json
{
"$schema": "https://opencode.ai/config.json",
"mcp": {
"sanity": {
"type": "remote",
"url": "https://mcp.sanity.io",
"oauth": {}
}
}
}
```
Then run: `opencode mcp auth sanity`
</details>
<details>
<summary><strong>Other clients</strong></summary>
For any MCP-compatible client, add `https://mcp.sanity.io` as the server URL.
If your client doesn't support remote MCP servers, use a proxy like `mcp-remote`:
```json
{
"mcpServers": {
"Sanity": {
"command": "npx",
"args": ["mcp-remote", "https://mcp.sanity.io", "--transport", "http-only"]
}
}
}
```
</details>
<br />
#### Authorization and troubleshooting
Manual MCP configuration uses OAuth by default. You can use token auth instead by setting an `Authorization: Bearer <token>` header in the MCP config. If authentication fails after CLI setup, rerun `npx sanity@latest mcp configure` and restart your MCP client. For OAuth reset issues, Cursor provides **Cursor: Clear All MCP Tokens** and VS Code provides **Authentication: Remove Dynamic Authentication Providers**.
See the [Sanity MCP docs](https://www.sanity.io/docs/ai/mcp-server) for authorization options and troubleshooting.
### Option 2: Install Agent Skills
Install best practices skills that work with any [Agent Skills](https://agentskills.io)-compatible agent.
```bash
npx skills add sanity-io/agent-toolkit
```
See [Option 3](#option-3-install-plugin) for plugin installation.
### Option 3: Install plugin
Install the Sanity plugin to get MCP server, agent skills, and commands.
Available on the [Claude Code marketplace](https://claude.com/plugins/sanity), [Cursor Marketplace](https://cursor.com/marketplace/sanity), or from this repo as an [Agent Plugin](https://agent-plugins.org/).
#### Claude Code
The Sanity plugin is listed on the [official Anthropic plugin marketplace](https://claude.com/plugins/sanity). The official marketplace (`claude-plugins-official`) is pre-registered when you start Claude Code — you do not need to add a custom marketplace.
Install from Claude Code:
```
/plugin install sanity@claude-plugins-official
```
If the plugin is not found, refresh the marketplace catalog and retry:
```
/plugin marketplace update claude-plugins-official
```
Then run `/reload-plugins` to activate without restarting.
**Alternative: interactive install**
1. Run `/plugin` and open the **Discover** tab
2. Search for **Sanity**
3. Review what the plugin will install — commands, skills, hooks, and MCP servers — before confirming ([Anthropic recommends reviewing plugin permissions and source before installing](https://code.claude.com/docs/en/discover-plugins#install-plugins))
4. Choose an installation scope:
- **User** (default): all projects on this machine
- **Project**: shared with collaborators via `.claude/settings.json`
- **Local**: this repository only
5. Run `/reload-plugins` to activate without restarting
**Verify installation:** Ask Claude Code: "which skills do you have access to?"
You should see the Sanity skills listed.
**Start using:** Use natural language and skills activate automatically:
> Help me create a blog post schema in Sanity
> Review my GROQ query and Next.js Visual Editing setup
Or run `/sanity` to explore all capabilities.
#### Cursor
Install from the [Cursor Marketplace](https://cursor.com/marketplace/sanity) by running this in Cursor chat:
```
/add-plugin sanity
```
**Verify installation:** Ask Cursor: "which skills do you have access to?"
You should see the Sanity skills listed.
**Start using:** Use natural language and skills activate automatically:
> Help me create a blog post schema in Sanity
> Review my GROQ query and Next.js Visual Editing setup
#### Codex
1. Add the Sanity marketplace:
```bash
codex plugin marketplace add sanity-io/agent-toolkit
```
2. Install the plugin from Codex's plugin directory (select the **Sanity Agent Toolkit** marketplace, then install **Sanity**).
3. Restart Codex. Verify by asking: "which skills do you have access to?" — you should see the Sanity skills listed.
### Option 4: Manual installation
Install the skill references locally to teach your editor Sanity best practices:
1. Copy `skills/sanity-best-practices/` to your project.
2. (Recommended) Copy `AGENTS.md` to your project root to act as a knowledge router.
---
## Capabilities
### MCP tools
With MCP connected, your AI can use tools like:
- `query_documents` — run GROQ queries directly
- `create_documents` — create draft documents from structured content, or version documents when a release ID is provided
- `patch_documents` — surgical edits to existing documents; published documents are edited by creating or updating drafts
- `publish_documents` / `unpublish_documents` — manage document lifecycle
- `deploy_schema` / `get_schema` — deploy MCP-managed schemas and inspect deployed schemas
- `deploy_studio` — deploy a hosted Studio bound to an MCP-managed schema
- `create_release` / `list_releases` — create and inspect Content Releases
- `create_version` — create version documents for releases
- `generate_image` / `transform_image` — AI image generation and editing
- `whoami` — verify the authenticated Sanity user
- `get_project_studios` — list Studio applications linked to a project
- `search_docs` / `read_docs` — search and read Sanity documentation
- `list_sanity_rules` / `get_sanity_rules` — load agent rules on demand
- `give_feedback` — report MCP tool errors, missing capabilities, confusing output, or documentation issues
MCP-managed schemas are resolved before Studio-deployed and legacy schemas. If you deploy schema changes with `deploy_schema`, redeploy any matching MCP-managed Studio with `deploy_studio` so it picks up the latest schema. `generate_image`, `transform_image`, and `create_version` with an `instruction` consume Sanity AI credits.
See the [full list of available tools](https://www.sanity.io/docs/ai/mcp-server#available-tools).
### Agent skills
Best practices skills that agents like Claude Code, Cursor, GitHub Copilot, etc. can discover and use automatically. Skills follow the [Agent Skills](https://agentskills.io) format. See [Option 2](#option-2-install-agent-skills) for installation.
| Skill | Description |
| :--- | :--- |
| **sanity-best-practices** | GROQ performance, schema design, Visual Editing, images, Portable Text, Studio, TypeGen, localization, migrations, and framework integration guides |
| **content-modeling-best-practices** | Structured content principles: separation of concerns, references vs embedding, content reuse |
| **seo-aeo-best-practices** | SEO/AEO with EEAT principles, structured data (JSON-LD), technical SEO patterns |
| **content-experimentation-best-practices** | A/B testing methodology, statistical foundations, experiment design |
| **sanity-singletons** | First-class singleton documents via `document.singletons`, Structure helpers, and migrating from userland or plugin-based singletons |
| **sanity-form-component-suspense** | Fix custom Studio form components (inputs, fields, array items) that suspend without their own `Suspense` boundary, causing layout shift or blocking new documents |
### Getting started flow
The onboarding guide follows three phases:
1. **Studio & Schema** — Set up Sanity Studio and define your content model
2. **Content** — Import existing content or generate placeholder content via MCP
3. **Frontend** — Integrate with your application (framework-specific)
Just say: "Get started with Sanity" to begin.
### Slash commands (Claude Code)
| Command | What it does |
| :--- | :--- |
| `/sanity` | List available skills and help topics |
| `/sanity-review` | Review code for Sanity best practices |
| `/typegen` | Run TypeGen and troubleshoot issues |
| `/deploy-schema` | Deploy schema with verification |
---
## Repository structure
> **Note:** The reference files in `skills/sanity-best-practices/references/` are the canonical content for the Sanity MCP server's `list_sanity_rules` / `get_sanity_rules` tools. Each file must have valid `name` and `description` frontmatter — rule names are derived from filenames (e.g., `nextjs.md` → `nextjs`).
```text
sanity-io/agent-toolkit/
├── AGENTS.md # Knowledge router & agent behavior
├── README.md # This file
├── plugin.json # Portable Agent Plugins v1 manifest
├── mcp.json # Portable Agent Plugins v1 MCP configuration
├── .agents/plugins/ # Codex marketplace
│ └── marketplace.json # Codex marketplace metadata
├── .claude-plugin/ # Claude Code plugin configuration (distributed via claude-plugins-official)
│ ├── plugin.json # Plugin manifest (name: sanity)
│ └── marketplace.json # Marketplace manifest for repo-based discovery
├── .codex-plugin/ # Codex plugin configuration
│ └── plugin.json # Codex plugin manifest
├── .cursor-plugin/ # Cursor plugin configuration (distributed via cursor.com/marketplace)
│ ├── marketplace.json # Cursor marketplace metadata
│ └── plugin.json # Per-plugin manifest
├── .mcp.json # Legacy client MCP compatibility configuration
├── assets/ # Plugin branding
│ └── logo.svg # Sanity logo for marketplace display
├── commands/ # Agent commands
│ ├── sanity.md # /sanity help
│ ├── sanity-review.md # /sanity-review
│ ├── typegen.md # /typegen
│ └── deploy-schema.md # /deploy-schema
├── scripts/ # Validation and CI scripts
│ ├── validate-agent-plugin.mjs # Validate manifests against the v1 JSON Schemas
│ └── validate-cursor-plugin.mjs # Cursor plugin validator
└── skills/ # Agent skills (agentskills.io format)
├── sanity-best-practices/ # Comprehensive Sanity skill
│ ├── SKILL.md # Skill definition and quick reference
│ └── references/ # Canonical content (25 guides)
│ ├── get-started.md # Onboarding guide
│ ├── nextjs.md # Next.js integration
│ ├── groq.md # GROQ patterns & performance
│ ├── schema.md # Schema design & validation
│ └── ... # See SKILL.md for full index
├── content-modeling-best-practices/ # Modeling guidance + topic references
├── seo-aeo-best-practices/ # SEO/AEO guidance + topic references
├── content-experimentation-best-practices/ # Experiment design + stats references
├── sanity-singletons/ # First-class singleton documents (Studio 6.17+)
└── sanity-form-component-suspense/ # Suspense boundaries for custom form components
```
All skills use `references/` for detailed content loaded on demand. The `sanity-best-practices` references are also the canonical source for the MCP server's Sanity rules.
---
## Resources
- [Create Sanity account](https://www.sanity.io/get-started)
- [Sanity documentation](https://www.sanity.io/docs)
- [GROQ language reference](https://www.sanity.io/docs/groq)
- [Visual Editing guide](https://www.sanity.io/docs/visual-editing)
- [Sanity TypeGen](https://www.sanity.io/docs/sanity-typegen)
- [MCP server docs](https://www.sanity.io/docs/ai/mcp-server)
- [Agent Plugins specification](https://agent-plugins.org/specification)
- [Blueprints Infrastructure as Code](https://www.sanity.io/docs/compute-and-ai/blueprints)
---
## Contributing
Found a better pattern? Missing a framework or best practice? Read the [contributing guide](CONTRIBUTING.md) for how skills work and what makes a good contribution, then:
1. Fork the repo.
2. Install dependencies with `npm install`.
3. Make your changes in `skills/<skill-name>/`.
4. Run `npm run validate:all` to check skill and plugin validity.
5. Submit a PR.
---
## Support
- [Sanity Community (Discord)](https://www.sanity.io/community/join)
- [GitHub issues](https://github.com/sanity-io/agent-toolkit/issues)
---
**License:** MIT