Simplenote MCP Server

A lightweight MCP server that integrates Simplenote with Claude Desktop using the MCP Python SDK.

This allows Claude Desktop to interact with your Simplenote notes as a memory backend or content source.

What's New

30 Tools — Full Bear Parity + Simplenote Differentiators + Claude Companion Tools + Vault Encryption

Vault — opt-in client-side note encryption: Simplenote has no encryption at rest. create_note/update_note now accept encrypt: true, and encrypt_note/decrypt_note convert existing notes — bodies become AES-256-GCM ciphertext before they ever reach Simplenote's API. See docs/security/encryption-design.md.

MCP Resources and Prompts hardened for the working-memory companion use case:

• Fixed: list_resources/read_resource were silently dropping tag/date/pagination metadata via non-schema fields — now attached through the MCP spec's _meta extension field, the correct mechanism.

• session-handoff MCP Prompt: scaffolds the Session Continuity workflow (get_or_create_note + add_text with a Status:/Next:/Blockers: format) for cross-session context handoff.

Irreversible-deletion tools with mandatory safety guards:

• permanent_delete_note: Permanently destroy a single note; requires confirm=true; dry-run preview by default

• empty_trash: Permanently delete all trashed notes; defaults to dry_run=true (preview); requires dry_run=false AND confirm=true

• 1334 tests passing, 79%+ coverage, zero linting/type errors

See the CHANGELOG and ROADMAP.md for complete details.

v1.17.0

• search_notes async fix: Boolean AND queries no longer hang the server; search now runs in a thread-pool executor with a 30 s timeout

• Substring pre-filter: searching "test" now correctly returns notes containing "testing", "tested", etc.

• Real-engine integration test suite added; import error in test helpers fixed

v1.16.0

• publish_note: Publish a note to a public URL — unique to Simplenote MCP; returns public_url

• unpublish_note: Remove a note from public access; no-op if already unpublished

See the CHANGELOG for complete details.

🔧 Features

• 📝 Full Note Management: Read, create, update, and delete Simplenote notes

• 🔍 Advanced Search: Boolean operators, phrase matching, tag and date filters

• ⚡ High Performance: In-memory caching with background synchronization

• 🔐 Secure Authentication: Token-based authentication via environment variables

• 🔑 Vault Encryption: Opt-in client-side AES-256-GCM encryption for sensitive notes — Simplenote itself has no encryption at rest

• 🧩 MCP Compatible: Works with Claude Desktop and other MCP clients

• 🐳 Docker Ready: Full containerization with multi-stage builds and security hardening

• 📊 Monitoring: Optional HTTP endpoints for health, readiness, and metrics

• 🧪 Robust Testing: Comprehensive test suite with 1334 tests and continuous integration

• 🔒 Security Hardened: Regular security scanning with Bandit, pip-audit, and dependency checks

🚀 Quick Start

Prerequisites

• Simplenote account (create one at simplenote.com)

• Python 3.10+ (for non-Docker installs) or Docker

Option 1: Docker (Recommended)

The fastest way to get started is using our pre-built Docker image:

# Pull and run the latest image
docker run -d \
  --name simplenote-mcp \
  -e SIMPLENOTE_EMAIL=your.email@example.com \
  -e SIMPLENOTE_PASSWORD=your-password \
  -e MCP_TRANSPORT=http \
  -e MCP_HTTP_HOST=0.0.0.0 \
  -e MCP_HTTP_AUTH_TOKEN=your-random-secret-token \
  -p 8000:8000 \
  docdyhr/simplenote-mcp-server:latest

MCP_HTTP_AUTH_TOKEN is required whenever MCP_HTTP_HOST is anything other than 127.0.0.1/localhost — the server refuses to start otherwise (see the Security section below). Without MCP_TRANSPORT=http, the server runs over stdio by default and nothing listens on the published port at all.

Docker Health Checks: health monitoring is a separate HTTP endpoint from the MCP protocol port above — it's off by default and must be enabled explicitly with -e ENABLE_HTTP_ENDPOINT=true -e HTTP_HOST=0.0.0.0 -p 8080:8080 (Docker's -p mapping forwards to the container's network interface, not its loopback, so HTTP_HOST must be 0.0.0.0 for the published port to actually reach it — the 127.0.0.1 default only works if you're calling these endpoints from another process inside the same container):

• Health: http://localhost:8080/health

• Readiness: http://localhost:8080/ready

• Metrics: http://localhost:8080/metrics (Prometheus format)

The server refuses to start if HTTP_HOST is non-loopback and no HTTP_ENDPOINT_AUTH_TOKEN is set, since these endpoints would otherwise be reachable by anyone who can reach the port. Set a bearer token (checked via Authorization: Bearer <token>, same mechanism as MCP_HTTP_AUTH_TOKEN above) if you need a non-loopback bind — loopback callers are always trusted regardless, so this never breaks a local health check. Prefer keeping it loopback-only and publishing with -p 127.0.0.1:8080:8080 instead of -p 8080:8080 when you can.

Or use Docker Compose:

# Clone the repository for docker-compose.yml
git clone https://github.com/docdyhr/simplenote-mcp-server.git
cd simplenote-mcp-server

# Set environment variables
export SIMPLENOTE_EMAIL=your.email@example.com
export SIMPLENOTE_PASSWORD=your-password

# Run with Docker Compose
docker-compose up -d

Option 2: Smithery (One-click install)

Install automatically via Smithery:

npx -y @smithery/cli install @docdyhr/simplenote-mcp-server --client claude

This method automatically configures Claude Desktop with the MCP server.

Option 3: Traditional Python Install

git clone https://github.com/docdyhr/simplenote-mcp-server.git
cd simplenote-mcp-server
pip install -e .
simplenote-mcp-server

🗂 Documentation Map & Archives

• Start with docs/DOCUMENTATION_GUIDE.md for a curated tour of user, developer, and operations docs plus maintenance checklists.

• Historical project summaries now live under docs/archive/2025/, keeping the repository root focused on active roadmaps and guides.

• Need something fast? Run rg "<topic>" docs/ or jump to docs/index.md for the MkDocs-style table of contents.

🐳 Docker Deployment

Container Features

• Multi-stage builds for optimized image size

• Security hardening with non-root user and minimal attack surface

• Health monitoring endpoints built-in

• Resource limits and proper signal handling

• Volume support for persistent data

Using Pre-built Images

The easiest way to use the server is with our pre-built Docker images:

# Pull the latest image
docker pull docdyhr/simplenote-mcp-server:latest

# Run with Docker (see Quick Start above for the required MCP_HTTP_* env vars)
docker run -d \
  -e SIMPLENOTE_EMAIL=your.email@example.com \
  -e SIMPLENOTE_PASSWORD=your-password \
  -e MCP_TRANSPORT=http \
  -e MCP_HTTP_HOST=0.0.0.0 \
  -e MCP_HTTP_AUTH_TOKEN=your-random-secret-token \
  -p 8000:8000 \
  docdyhr/simplenote-mcp-server:latest

# Or use Docker Compose (set MCP_HTTP_AUTH_TOKEN in your environment/.env first)
docker-compose up -d

Available tags:

• latest - Latest stable release

• v1.18.0 - Specific version

• main - Latest development build

Production Deployment

# Build and run the production container
docker-compose up -d

# Or build manually
docker build -t simplenote-mcp-server .
docker run -d \
  -e SIMPLENOTE_EMAIL=your.email@example.com \
  -e SIMPLENOTE_PASSWORD=your-password \
  -e MCP_TRANSPORT=http \
  -e MCP_HTTP_HOST=0.0.0.0 \
  -e MCP_HTTP_AUTH_TOKEN=your-random-secret-token \
  -p 8000:8000 \
  simplenote-mcp-server

Development with Docker

# Use the development compose file for live code mounting
docker-compose -f docker-compose.dev.yml up

Docker Features

• Multi-stage build for optimized image size (346MB)

• Multi-platform support: linux/amd64 and linux/arm64

• Security hardening: Non-root user, read-only filesystem, no new privileges

• Health checks and automatic restart policies

• Resource limits: 1 CPU, 512MB memory

• Logging: Persistent log volumes

• Environment-based configuration

• CI/CD Pipeline: Automated builds and publishing to Docker Hub

• Security scanning: Trivy vulnerability scanning on all images

• Container signing: Sigstore cosign signatures for supply chain security

• Kubernetes ready: Production-grade Helm chart with security hardening

• Automated updates: Dependabot for dependencies, auto-versioning workflows

• Health monitoring: Continuous health checks and alerting

• Enterprise notifications: Slack and email integration for CI/CD status

☸️ Kubernetes Deployment

Using Helm (Recommended)

Deploy to Kubernetes with our production-ready Helm chart:

# Install from local chart
helm install my-simplenote ./helm/simplenote-mcp-server \
  --set simplenote.email="your-email@example.com" \
  --set simplenote.password="your-password"

# Or with external secrets (recommended for production)
helm install my-simplenote ./helm/simplenote-mcp-server \
  --set externalSecrets.enabled=true \
  --set externalSecrets.secretStore.name="vault-backend"

Kubernetes Features

• Security hardening: Non-root user, read-only filesystem, dropped capabilities

• Resource management: CPU/memory limits and requests configured

• Auto-scaling: Horizontal Pod Autoscaler support

• Health checks: Liveness and readiness probes

• External secrets: Integration with external secret management

• Service mesh ready: Compatible with Istio and other service meshes

Production Configuration

# values.yaml for production
replicaCount: 3
autoscaling:
  enabled: true
  minReplicas: 2
  maxReplicas: 10
resources:
  limits:
    cpu: 1000m
    memory: 512Mi
  requests:
    cpu: 500m
    memory: 256Mi

⚙️ Configuration

Environment Variables

Claude Desktop Integration

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "simplenote": {
      "description": "Access and manage your Simplenote notes",
      "command": "simplenote-mcp-server",
      "env": {
        "SIMPLENOTE_EMAIL": "your.email@example.com",
        "SIMPLENOTE_PASSWORD": "your-password",
        "CACHE_MAX_SIZE": "10000"
      }
    }
  }
}

🔍 Advanced Search

Powerful search with boolean logic and filters:

# Boolean operators
project AND meeting AND NOT cancelled

# Phrase matching
"action items" AND project

# Tag filtering
meeting tag:work tag:important

# Date ranges
project from:2023-01-01 to:2023-12-31

# Combined query
"status update" AND project tag:work from:2023-01-01 NOT cancelled

🛠️ Available Tools

📊 Performance & Caching

• In-memory caching with background synchronization

• Pagination support for large note collections

• Indexed lookups for tags and content

• Query result caching for repeated searches

• Optimized API usage with minimal Simplenote calls

🎯 Recent Improvements

✅ January 2025 - Performance & Code Quality

Critical Bug Fix:

• Fixed Claude Desktop timeout - Reduced startup time from 55+ seconds to < 1 second (98% improvement)

• Implemented thread pool execution for blocking Simplenote API calls

• Made cache initialization truly non-blocking with background loading

• Resolved anyio.BrokenResourceError during shutdown

Code Refactoring - Phase 1 Complete:

• Cache module complexity reduced: 5 high-complexity functions (CC >= 15) → 0 (100% reduction)

• Maintainability improved: Cache MI from 12.7 → 16.2 (+28%)

• Extracted 23 helper methods for better code organization

• All 670 tests passing with 67% cache coverage maintained

• See REFACTORING_PHASE1_COMPLETE.md for details

Documentation Enhancements:

• Added comprehensive CHANGELOG.md with complete version history

• Created TESTING_CLAUDE_DESKTOP.md for user testing guide

• Added code complexity analysis tools (check_complexity.py)

• Documented refactoring plan and completion reports

Quality Tools:

• Integrated Radon for automated complexity analysis

• Baseline metrics: 22 functions CC >= 15 (down from 28)

• Average Maintainability Index: 57.9 (maintained)

• Zero diagnostics errors, all quality gates passing

✅ September 2025 - Quality & Reliability Enhancements

✅ Quality & Reliability Enhancements

Test Suite Stabilization:

• Fixed test isolation issues that caused intermittent failures

• Improved test cleanup with proper timeout handling

• Enhanced fixture management for better test reliability

• Achieved consistent test results across individual and suite runs

CI/CD Pipeline Optimization:

• Consolidated 28 workflows down to 16 active workflows

• Implemented unified monitoring workflow combining security, health, and badge checks

• Improved test coverage reporting with realistic 15.6% baseline

• Enhanced Docker build validation and security scanning

Code Quality Improvements:

• All linting (Ruff), formatting, and type checking (MyPy) now pass consistently

• Zero high-severity security vulnerabilities (verified with Bandit, pip-audit, safety)

• Standardized code formatting and pre-commit hooks configuration

• Enhanced error handling and user-facing error messages

🔧 Developer Experience

Improved Testing:

• 724 comprehensive tests covering core functionality

• Function-scoped fixtures for better test isolation

• Realistic coverage baseline established (15.6%)

• Streamlined test execution with proper cleanup

Enhanced Documentation:

• Updated deployment guides with current Docker setup

• Improved health monitoring endpoint documentation

• Added troubleshooting guides for common issues

• Current status and roadmap documentation

Container Improvements:

• Multi-stage Docker builds for optimized image size

• Built-in health monitoring endpoints (/health, /ready, /metrics)

• Enhanced security hardening with non-root user

• Improved signal handling and graceful shutdown

🧪 Testing & Evaluation

MCP Evaluations ✅

Status: ✅ WORKING - Complete mcp-evals integration with TypeScript wrapper!

This project includes comprehensive evaluations using mcp-evals to ensure reliability and performance:

# Setup evaluation environment
npm install
npm run validate:evals

# Run evaluation suites
npm run eval:smoke          # Quick smoke tests (2-3 minutes) ✅ VERIFIED
npm run eval:basic          # Standard evaluations (5-10 minutes)
npm run eval:comprehensive  # Full evaluation suite (15-30 minutes)

Latest Test Results: 4/5 tests passing excellently (avg 4.1/5):

• Server Startup: 4.6/5 ⭐ (Excellent)

• Authentication: 4.0/5 ⭐ (Good)

• Note Operations: 3.8/5 ⭐ (Good)

• Search: 5.0/5 ⭐ (Perfect)

• Error Handling: 1.4/5 ⚠️ (Needs improvement)

Evaluation Types

• Smoke Tests: Basic functionality validation

• CRUD Operations: Note creation, reading, updating, deletion

• Search & Filtering: Boolean search, tag filtering, date ranges

• Error Handling: Authentication, network issues, edge cases

• Performance: Large datasets, concurrent operations

• Security: Input validation, authentication enforcement

Automated Testing

Evaluations run automatically on:

• Pull Requests: Smoke + basic tests

• Releases: Comprehensive evaluation suite

• Manual Trigger: Full test matrix with detailed reporting

The evaluations use OpenAI's GPT models to assess:

• Accuracy: Correctness of responses

• Completeness: Thoroughness of results

• Relevance: Response appropriateness

• Clarity: Response readability

• Performance: Operation efficiency

📁 See evals/README.md for detailed evaluation documentation.

Traditional Testing

# Python unit tests
pytest

# Code quality checks
ruff check .
mypy simplenote_mcp

🛡️ Security

• Token-based authentication via environment variables

• No hardcoded credentials in Docker images

• Security-hardened containers with non-root users

• Read-only filesystem in production containers

• Resource limits to prevent abuse

• MCP HTTP transport is fail-closed by default: MCP_TRANSPORT=http refuses to start on any non-loopback MCP_HTTP_HOST unless MCP_HTTP_AUTH_TOKEN is set (a shared bearer secret, checked via constant-time comparison). Loopback binds (127.0.0.1/localhost) work without a token, matching stdio's local-process trust level. Set MCP_HTTP_ALLOWED_HOSTS/MCP_HTTP_ALLOWED_ORIGINS (comma-separated) to enable DNS-rebinding protection for non-loopback binds. This is intended for private networks (behind a VPN/Tailscale/SSH tunnel) — a static shared token has none of OAuth's revocation/audit/expiry properties, so avoid exposing it directly to the public internet even with a token set.

🚨 Troubleshooting

Common Issues

Authentication Problems:

• Verify SIMPLENOTE_EMAIL and SIMPLENOTE_PASSWORD are set correctly

• Check for typos in credentials

Docker Issues:

# Check container logs
docker-compose logs

# Restart services
docker-compose restart

# Rebuild if needed
docker-compose up --build

Claude Desktop Connection:

# Verify tools are available
./simplenote_mcp/scripts/verify_tools.sh

# Monitor logs
./simplenote_mcp/scripts/watch_logs.sh

Diagnostic Commands

# Test connectivity
python simplenote_mcp/tests/test_mcp_client.py

# Check server status
./simplenote_mcp/scripts/check_server_pid.sh

# Clean up and restart
./simplenote_mcp/scripts/cleanup_servers.sh

📚 Development

Quick Setup with mcp-evals

# One-command setup including evaluations
./setup-dev-env-with-evals.sh

# Or manual setup
git clone https://github.com/docdyhr/simplenote-mcp-server.git
cd simplenote-mcp-server
pip install -e ".[dev,test]"
npm install  # For mcp-evals

Local Development

# Run the server
python simplenote_mcp_server.py

# Run Python tests
pytest

# Run mcp-evals
npm run eval:smoke    # Quick validation
npm run eval:basic    # Standard tests
npm run eval:all      # Full test suite

# Code quality
ruff check .
ruff format .
mypy simplenote_mcp

Development Environment

The setup script creates:

• Python development environment with all dependencies

• Node.js environment for mcp-evals

• Example configuration files

• Pre-commit hooks

• Validation for all evaluation files

Testing Strategy

1. Unit Tests: Traditional Python pytest for core logic

2. Integration Tests: MCP protocol compliance testing

3. Smoke Tests: Quick validation of basic functionality

4. Evaluation Tests: LLM-based assessment of real-world usage

5. Performance Tests: Load and stress testing

Running MCP Evaluations

Docker Method (Recommended)

Due to potential permission issues with tsx, we recommend running MCP evaluations in Docker:

# Run smoke tests
./scripts/run-evals-docker.sh smoke

# Run basic evaluations
./scripts/run-evals-docker.sh basic

# Run comprehensive evaluations
./scripts/run-evals-docker.sh comprehensive

# Run all evaluations
./scripts/run-evals-docker.sh all

Direct Method (if permissions allow)

npm run eval:smoke
npm run eval:basic
npm run eval:comprehensive
npm run eval:all

Docker Development

# Development with live code reload
docker-compose -f docker-compose.dev.yml up

# Build and test
docker build -t simplenote-mcp-server:test .
docker run --rm simplenote-mcp-server:test --help

🤝 Contributing

Contributions are welcome! Please read CONTRIBUTING.md for guidelines.

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🔗 Related Projects

• Model Context Protocol

• MCP Example Servers

⭐ Support the Project

If you find this project helpful, please consider giving it a star on GitHub! Your support helps:

• 🚀 Increase visibility for other developers who might benefit from this tool

• 💪 Motivate continued development and maintenance

• 📈 Build community around the Model Context Protocol ecosystem

• 🛡️ Validate trust through community engagement

⭐ Star this repository — it takes just one click and means a lot!