Grammar checking, AI content detection, paraphrasing, and the rest of Sapling's language tools — available directly inside Claude, Claude Code, and any other MCP-compatible AI assistant.
https://mcp.sapling.ai/mcpclaude mcp add --transport http sapling \
https://mcp.sapling.ai/mcpThe Model Context Protocol is an open standard that lets AI assistants call external tools and APIs in a structured way. The Sapling MCP Server exposes Sapling's language tools as MCP tools — so you can ask your assistant to grammar-check a document, detect AI-generated content, paraphrase a sentence, or analyze sentiment without leaving the conversation.
Building or using a browser-based AI agent instead? Sapling's free tool pages also support WebMCP, so a supporting agent can call them directly with no server or API key involved.
"Check this email for grammar errors." Your assistant decides a Sapling tool is the right way to answer.
Connect to the hosted server at https://mcp.sapling.ai/mcp, or have your client start @saplingai/mcp-server locally over stdio.
The tool calls the Sapling API and returns structured results — edits, scores, and suggestions — back into the conversation.
Tools take text (sapling_ai_detect_url takes a web page’s url, and autocomplete takes a short query) and return structured JSON your assistant can reason over. The hosted server has 22 tools. Every version of the local package includes the seven below, and newer releases add more.
sapling_grammar_check
sapling_spell_check
Catch grammar errors, spelling mistakes, punctuation issues, and style problems in any text. Returns edit suggestions with types (preposition, verb tense, etc.) and confidence scores.
sapling_ai_detect
Detect whether text was written by a human or generated by GPT-5, Claude 4.5, Gemini 2.5, Qwen3, DeepSeek-V3, or other AI models. Returns an overall score plus per-sentence predictions.
sapling_rephrase
Rewrite text with different tone, formality level, or style while preserving meaning. Useful for adapting content for different audiences.
sapling_summarize
Generate concise summaries of longer documents, emails, or articles.
sapling_sentiment
Classify text as positive, negative, or neutral. Useful for reviewing customer feedback, tickets, or social media posts within an AI workflow.
sapling_language_detect
Identify the language of a text snippet. Supports 30+ languages including English, Spanish, French, German, Chinese, Japanese, Arabic, and more.
Also on the hosted server:
sapling_ai_detect_url
sapling_tone
sapling_translate
sapling_detect_pii
sapling_ner
sapling_classify
sapling_extract
sapling_simplify
sapling_quality
sapling_readability_statistics
sapling_seo
sapling_styleguide
sapling_inclusive
sapling_safety
sapling_autocomplete
On the hosted server, spell check and language detection are named sapling_spellcheck and sapling_detect_language. The full tool reference lists which tools each local package version includes.
Three steps, about two minutes. See the complete setup guide for other MCP clients and troubleshooting.
You need a Sapling API key; trial keys are available right after registration. The hosted server needs nothing installed. The local server needs Node.js 18 or newer (including npx). Keep keys in your client's configuration — not in prompts or committed files.
Hosted server
Add Sapling as a remote MCP server (in claude.ai and Claude Desktop, a custom connector) with these settings. On first connection your client opens a Sapling sign-in page where you choose which API key the connection uses.
| Server URL | https://mcp.sapling.ai/mcp |
|---|---|
| Transport | Streamable HTTP |
| Auth | OAuth (sign in with Sapling), or the header Authorization: Bearer YOUR_API_KEY |
| Server card | /.well-known/mcp/server-card.json, for MCP catalogs and registries |
In Claude Code:
claude mcp add --transport http sapling \
https://mcp.sapling.ai/mcpThen run /mcp in Claude Code and choose Authenticate to sign in.
Claude Code
Run in your terminal, replacing YOUR_API_KEY with your Sapling API key:
claude mcp add sapling --env SAPLING_API_KEY=YOUR_API_KEY \
-- npx -y @saplingai/mcp-serverThen run claude mcp list to confirm sapling is configured.
Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop:
{
"mcpServers": {
"sapling": {
"command": "npx",
"args": ["-y", "@saplingai/mcp-server"],
"env": {
"SAPLING_API_KEY": "YOUR_API_KEY"
}
}
}
}Other MCP clients
Any client that accepts a command, arguments, and environment variables can run the server:
| Transport | stdio |
|---|---|
| Command | npx |
| Arguments | -y, @saplingai/mcp-server |
| Environment | SAPLING_API_KEY=YOUR_API_KEY |
Your client starts the local process, which then calls https://api.sapling.ai. For a remote endpoint, use the hosted server at https://mcp.sapling.ai/mcp.
Claude Code asks for approval the first time each Sapling tool runs. To pre-approve them — or if a call is denied instead of prompting — run /permissions and add an allow rule for mcp__sapling (all Sapling tools) or a single tool such as mcp__sapling__sapling_grammar_check. See tool permissions for details.
The package doesn't log or persist your key or your text. Optional no-data-retention mode, plus on-premises deployment for HIPAA and PCI compliance.
Sapling's API is optimized for real-time use. Grammar checks typically return in under 200 ms, so tool calls don't stall the conversation.
Grammar checking and language detection across 30+ languages including Chinese, German, Spanish, Japanese, and more.
Any MCP client that supports remote servers, including claude.ai, Claude Desktop and Claude Code, can connect to the hosted server at https://mcp.sapling.ai/mcp. Clients that only support local stdio servers can run the npm package instead, with command npx, arguments -y @saplingai/mcp-server, and SAPLING_API_KEY in the environment.
https://mcp.sapling.ai/mcp, over Streamable HTTP. It accepts OAuth sign-in with a Sapling account or a Sapling API key sent as Authorization: Bearer YOUR_API_KEY. MCP catalogs that read server cards can use https://sapling.ai/.well-known/mcp/server-card.json.
Nothing for the hosted server. The local server needs Node.js 18 or newer, including the npx command, and runs on your machine, calling the Sapling API over HTTPS.
Yes. With the hosted server you sign in to Sapling and choose which of your API keys the connection uses, or send a key in the Authorization header. With the local server, set your key as the SAPLING_API_KEY environment variable. Trial keys are available after registration.
The MCP package does not log or persist your API key or submitted text. Tool inputs are sent to the configured Sapling API endpoint for processing. Contact us if you require no-data-retention or an on-premises deployment.
Yes. Set SAPLING_API_BASE_URL to your own Sapling endpoint. Only point it at an endpoint you trust — your API key and submitted text are sent there.
Grab an API key, add https://mcp.sapling.ai/mcp to your client, and start grammar-checking, detecting, and rephrasing from inside the tools you already use.
Need volume pricing, no-data-retention, or on-premises? Talk to us.