DEV.co
MCP Servers · alioshr

memory-bank-mcp

Memory Bank MCP is a TypeScript server that provides remote access to AI assistant memory banks through the Model Context Protocol. It enables multi-project memory management with file-level isolation, compatible with Claude, Cline, and Cursor.

Source: GitHub — github.com/alioshr/memory-bank-mcp
913
GitHub stars
84
Forks
TypeScript
Primary language
MIT
License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

FieldValue
Repositoryalioshr/memory-bank-mcp
Owneralioshr
Primary languageTypeScript
LicenseMIT — OSI-approved
Stars913
Forks84
Open issues17
Latest releaseUnknown
Last updated2025-08-20
Sourcehttps://github.com/alioshr/memory-bank-mcp

What memory-bank-mcp is

A Node.js/TypeScript MCP server implementing read/write/update operations for memory bank files with project-based path traversal protection and type-safe operations. Deployable via npm, Docker, or direct process invocation with environment-based configuration.

Quickstart

Get the memory-bank-mcp source

Clone the repository and explore it locally.

terminalbash
git clone https://github.com/alioshr/memory-bank-mcp.gitcd memory-bank-mcp# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

Multi-project AI Assistant Memory Management

Teams using Claude, Cline, or Cursor across multiple codebases can centralize memory bank storage with per-project isolation, enabling consistent context sharing without file duplication.

Shared Development Team Context

Engineering teams can maintain a centralized memory bank server accessible to multiple AI assistants simultaneously, reducing context setup time and improving consistency across pair-programming sessions.

CI/CD-integrated AI Assistant Operations

Integrate memory banks into deployment pipelines or automation workflows where AI assistants need persistent, validated access to project-specific knowledge without manual configuration.

Implementation considerations

  • Set MEMORY_BANK_ROOT environment variable to a writable directory; incorrect paths will cause silent failures or permission errors.
  • Validate that the MCP server process has read/write access to the memory bank directory and that parent processes (Claude, Cursor, etc.) correctly pass stdio to the server.
  • Test path traversal prevention by attempting to access files outside the configured project directory; confirm isolation works as intended.
  • Plan for log rotation and disk usage if using Docker; the server may generate logs without built-in cleanup.
  • Ensure TypeScript compilation or npx caching is compatible with your deployment environment; network isolation may block npm package retrieval.

When to avoid it — and what to weigh

  • Single-user, local-only workflow — If you work alone on one project locally, file-based memory banks (native to Claude/Cline) are simpler; this adds operational overhead without benefit.
  • Sensitive data in memory banks — This project does not clearly document encryption, authentication, or audit logging. Requires careful review before storing confidential project data.
  • High-availability or mission-critical systems — No release history, versioning, or production deployment guidance provided. Unsuitable for systems requiring SLAs or formal rollback procedures.
  • AI assistants outside the Anthropic/Cursor ecosystem — Designed specifically for MCP-compatible clients (Claude, Cline, Cursor, Windsurf). Integration with other AI platforms is unknown.

License & commercial use

Licensed under MIT (permissive OSI license). Commercial use, modification, and distribution are permitted with attribution.

MIT license permits commercial use without restriction. However, verify that your deployment (hosting, modifications, support) complies with any internal commercial policies. No warranty or liability protection explicitly stated in the README; review LICENSE file for details.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

SignalAssessment
MaintenanceActive
DocumentationAdequate
License clarityClear
Deployment complexityLow
DEV.co fitGood
Assessment confidenceHigh
Security considerations

No authentication or authorization layer; any process with access to the MCP socket can read/write all project memory banks. Path traversal prevention is claimed but requires independent validation. No TLS, encryption-at-rest, or audit logging mentioned. Environment variables store sensitive paths; ensure process isolation and minimal privilege. Validate input validation and error messages do not leak filesystem paths.

Alternatives to consider

Anthropic Cline Memory Bank (file-based)

Native, no separate server process required. Use if single-user or isolated team workflow suffices.

Custom MCP server (self-built)

Full control over authentication, logging, and integrations. Higher development cost but tailored to security/compliance requirements.

Vector database (e.g., Pinecone, Weaviate) + embeddings

Semantic search and scalable context retrieval. Overkill for static memory banks but suitable for large knowledge bases.

Software development agency

Build on memory-bank-mcp with DEV.co software developers

Memory Bank MCP simplifies shared context across your team's AI tools. Review the security and deployment considerations above, test thoroughly in your environment, and deploy with confidence.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

Related on DEV.co

Explore the category and the services that help you build with it.

memory-bank-mcp FAQ

Can I use this in production without modifications?
Unknown. No production deployment guidance, SLA, or security audit documented. Review and test thoroughly, especially around authentication and isolation.
Does this work with OpenAI, Gemini, or other AI platforms?
No. This implements MCP server protocol, which is specific to Anthropic Claude, Cline, Cursor, and Windsurf. Other platforms require separate integration.
What happens if two clients try to write to the same memory file simultaneously?
Not documented. Likely filesystem-dependent (last-write-wins or race condition). No locking or conflict resolution mentioned.
Can I encrypt memory banks at rest?
Not built-in. You must encrypt the MEMORY_BANK_ROOT directory using OS-level tools (e.g., LUKS, BitLocker) or manage encryption outside this server.

Software developers & web developers for hire

DEV.co is a software development agency delivering custom software development services to companies building on open source. Our software developers and web developers design, integrate, and ship production systems — spanning web development, APIs, AI, data, and cloud. If memory-bank-mcp is part of your mcp servers roadmap, our team can implement, customize, migrate, and maintain it.

Ready to centralize AI assistant memory?

Memory Bank MCP simplifies shared context across your team's AI tools. Review the security and deployment considerations above, test thoroughly in your environment, and deploy with confidence.