AgentStack
MCP verified MIT Self-run

Claude Continuity

mcp-johnmwhitman-claude-continuity · by johnmwhitman

Archived: Early MCP memory experiment for Claude Desktop (2026). Superseded by Mem0 and others.

No reviews yet
0 installs
2 views
0.0% view→install

Install

$ agentstack add mcp-johnmwhitman-claude-continuity

✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

Are you the author of Claude Continuity? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Continuity

Persistent memory and context for Claude Desktop. Every conversation picks up where the last one left off.


Most AI assistants forget you the moment the session ends.

You paste the same context document every time. You re-explain your project. You remind Claude who your team is, what you're building, what you decided last week. Every conversation starts from zero.

Continuity fixes that.

It gives Claude Desktop a persistent memory layer — backed by ChromaDB running locally on your machine — so that what you share in one conversation is available in the next. Not uploaded to a server. Not summarized by someone else's model. Yours, on your hardware, under your control.


What it does

  • Remembers across sessions — decisions, projects, people, tasks, and context persist between conversations
  • Searches semantically — Claude can recall relevant memories using natural language, not exact keyword matching
  • Runs locally — ChromaDB lives on your machine. Nothing leaves unless you choose to share it.
  • Works autonomously — a nightly pipeline updates, prunes, and organizes memory without you thinking about it
  • MCP-native — built on the Model Context Protocol, the same standard Claude Desktop uses natively

What it's not

  • Not a plugin you install from a marketplace
  • Not a cloud service or SaaS product
  • Not a wrapper around someone else's memory API
  • Not finished — this is v0.1, built by one person, and it works

How it works

Your conversation
      ↓
Claude Desktop (with MCP)
      ↓
Continuity MCP server (memory_save / memory_search / memory_list)
      ↓
ChromaDB (local vector store, 6 collections)
      ↓
Next conversation — context already loaded

Six memory collections: projects, people, decisions, knowledge, conversations, tasks

Each entry is stored as a vector embedding, so Claude can find related memories even when the wording doesn't match exactly.


Requirements

  • Windows 10/11 (primary platform; Mac/Linux adaptable)
  • Claude Desktop with an active Anthropic subscription
  • Python 3.10+
  • ~500MB disk space for ChromaDB

Quick start

# 1. Clone the repo
git clone https://github.com/johnmwhitman/claude-continuity.git
cd claude-continuity

# 2. Install dependencies
pip install -r requirements.txt

# 3. Run the onboarding wizard
python automation/onboarding_wizard.py

# 4. Add the system prompt to Claude Desktop
# Settings → Profile → paste contents of docs/SYSTEM_PROMPT_TEMPLATE.md

# 5. Restart Claude Desktop

Full install guide: [docs/INSTALL.md](docs/INSTALL.md)


Project structure

claude-continuity/
├── automation/          # Core scripts — memory server, pipeline, utilities
├── docs/                # Install guide, architecture, system prompt template
├── web/                 # Dashboard and local web tools
├── .gitignore
├── CONTRIBUTING.md
├── LICENSE              # MIT
├── README.md
└── requirements.txt

Privacy

Your memories never leave your machine unless you explicitly push them somewhere. No telemetry. No analytics. No phone-home. The ChromaDB database lives at a local path you control.


Contributing

This project is early and welcomes help — especially on Mac/Linux compatibility, documentation, and testing. Read [CONTRIBUTING.md](CONTRIBUTING.md) before submitting a PR.

The one non-negotiable: no telemetry, no cloud dependencies, no data leaving the user's machine without explicit opt-in.


Status

v0.1 — working, opinionated, Windows-first.

The nightly pipeline runs. The memory server works. The onboarding wizard walks you through setup. It's not polished. It does what it says.


License

MIT © 2026 John Whitman


Built on the Elk River in southwest Missouri. Started because pasting the same context document into every Claude session got old.

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet — be the first.

Versions

  • v0.1.0 Imported from the upstream source.