Claustrum

English README | 한국어 README | 日本語 README | Español README | 中文 README

Claustrum is a shared memory layer for AI systems. It integrates context across projects, tools, and teams.

What This Project Does

• Shares memory context across multiple computers and developers.
• Keeps MCP workflows production-safe (stdout clean, policy-driven behavior).
• Provides team operations via Admin UI (workspace/project/user/permissions/audit).
• Supports external context sources (Notion, Jira, Confluence, Linear, Slack).

Why This Exists

AI coding context often gets fragmented:

• each machine has different memory state
• each teammate sees different context
• project decisions are lost in commits/chats/docs

Claustrum turns that into a shared, queryable memory system for teams.

Core Components

• Memory Core: REST API + policy + Postgres storage.
• MCP Adapter: stdio MCP bridge to Memory Core.
• Admin UI: management dashboard for teams.
• Shared Package: schemas/types/utilities used across apps.

Documentation (Pages-first)

This README is intentionally short. Detailed setup/config/operations are published on GitHub Pages and sourced from docs/content.

• Docs Site (GitHub Pages)
• One-line MCP config helper: curl -fsSL https://raw.githubusercontent.com/stephen-kim/claustrum/main/scripts/mcp-config-helper.js -o ./claustrum-mcp-config-helper.js && node ./claustrum-mcp-config-helper.js

Repository Layout

apps/
  docs-site/
  memory-core/
  mcp-adapter/
  admin-ui/
packages/
  shared/

Upstream

• Upstream (upstream): https://github.com/Intina47/context-sync.git