Jez | Jeremy Dawes
Back to projects
mcp active

Spell Checker MCP Server

Multi-language spell checking and AI-powered grammar correction for AI agents. 53 languages with DeepSeek R1 integration.

MCPCloudflare WorkersDeepSeek R1HunspellR2

Overview

A remote MCP server providing enterprise-grade spell checking and AI-powered grammar correction. Supports 53 languages with automatic detection, sub-50ms response times, and intelligent corrections using DeepSeek R1.

53
Languages
3
MCP Tools
<50ms
Response Time

The Problem

AI agents generating or processing text need reliable spell checking that:

  • Works across multiple languages without manual configuration
  • Provides precise error locations for targeted fixes
  • Distinguishes between spelling errors and grammar issues
  • Can automatically apply corrections when needed

Traditional spell checkers lack AI integration and multi-language auto-detection.

Solution

Three specialised tools for different use cases:

spell_check_analyze1

Error detection

spell_check_grammar1

AI grammar check

spell_check_correct1

Auto-correction

Tools

spell_check_analyze

Fast spelling error detection using Hunspell dictionaries:

  • Precise line and column positions for each error
  • Multiple correction suggestions per error
  • Language auto-detection or manual override
  • Sub-50ms response times (no AI involved)

spell_check_grammar

AI-powered grammar and style analysis using DeepSeek R1:

  • Grammar and punctuation error detection
  • Style improvement suggestions
  • Rule categorisation for each issue
  • Context-aware corrections

spell_check_correct

Automatic correction with document storage:

  • Fix spelling errors, grammar issues, or both
  • Corrected documents stored in R2
  • 30-day retention with retrieval URL
  • Preserves original formatting

Features

  • 53 Languages - Comprehensive Hunspell dictionary support
  • Auto-Detection - Identifies language automatically
  • DeepSeek R1 - 32B parameter model for grammar analysis
  • Edge Deployment - Global low-latency via Cloudflare Workers
  • OAuth Security - Google authentication for access control
  • R2 Storage - Corrected documents with 30-day retention

Architecture

Text Input → Language Detection → Hunspell Analysis

                              Spelling Errors + Positions

                    [Optional] DeepSeek R1 Grammar Check

                    [Optional] Auto-Correct → R2 Storage

Use Cases

Content Generation - Ensure AI-generated content is error-free

Document Processing - Batch spell check uploaded documents

Multi-Language Support - Handle international content without configuration

Quality Assurance - Add spell checking to content workflows

Interested in a similar solution?

Let's talk about your project