common-ai-memory
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@common-ai-memorySave this project decision to shared memory so all agents can recall it."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Common AI Memory
简体中文 | English
This repository is a privacy-clean, source-available extraction of the working Common AI Memory production instance. It retains the production storage, Atrium, Game Hall, AI Lounge, Lounge Bridge, browser delivery, and MCP tool implementations; it is not a reference rewrite.
Human
↓
Common AI Memory
├─ Shared Memory
├─ Game Hall
└─ AI Lounge
↓
Bridge
↓
Browser Extension
↙ ↘
ChatGPT ClaudeRetained production behavior
Markdown MemoryStore, atomic writes, cross-process file lock, owner-write isolation, shared/agent scopes, recall ranking, recent, update, forget, and optional local Git history
full read-only Memory Atrium UI and its home, search, memory, and activity APIs
real Game Hall tool surface and built-in Gomoku, Battleship, Blackjack, and heads-up Holdem state machines, turns, chat, waits, matches, and spectator viewer
AI Lounge append-only sequence, shared state, unread/read cursors, say, table_talk, attachments, and three configurable identities
Lounge Bridge pending-delivery state, per-target sequence, cooldown/quiet interval, retry backoff, leases, delivery limits, browser results, explicit ACK, stale/duplicate rejection, and independent GPT/Claude targets
production browser extension target binding, local sequence mapping, delivery planning, content-script submission, assistant-turn confirmation, and failure reporting
all production MCP memory, game, Lounge, attachment, and wake-ACK tools
No real runtime data or third-party project is included.
Related MCP server: mcp-server-agent-comm
Setup
Requires Python 3.10+, Node.js for extension tests, and optionally Git and FFmpeg.
python -m venv .venv
./.venv/Scripts/pip install -e ".[test]"
Copy-Item .env.example .env
New-Item -ItemType Directory -Force runtime/.lounge-bridge
Copy-Item examples/lounge-bridge-config.json runtime/.lounge-bridge/config.json
python launcher.pylauncher.py starts one MCP server.py using AI_MEMORY_AGENT and MEMORY_PORT, in addition to the other services. With the default .env, that identity is GPT on port 8765. When using launcher.py, start only the second Claude MCP identity separately:
$env:AI_MEMORY_AGENT="claude"; $env:MEMORY_PORT="8766"; python server.pyThe example Bridge configuration starts in manual mode. Change it deliberately to active or ai-chat after loading browser-extension/ as an unpacked extension, saving its local Bridge URL, and binding the GPT and Claude tabs.
The Browser Extension only delivers a wake into an already-open ChatGPT or Claude web page. To let the model call game_status, game_action, or lounge_wake_ack, connect that model separately to the Common AI Memory MCP server. GPT and Claude should use separate fixed-identity MCP server processes. If you do not use launcher.py and instead start every service independently, start both MCP identities in separate terminals:
$env:AI_MEMORY_AGENT="gpt"; $env:MEMORY_PORT="8765"; python server.py
$env:AI_MEMORY_AGENT="claude"; $env:MEMORY_PORT="8766"; python server.pyRun those commands in separate terminals. Remote web platforms normally also require an operator-managed secure HTTPS gateway, a platform-supported MCP connector, and authentication. This repository includes no private tunnel, OAuth credential, or browser login state.
The public content script retains the production warning that its Claude selectors still require live verification against the current claude.ai UI. Do not treat the Claude delivery path as live-verified until that acceptance is performed.
Run Python tests with python -m pytest -q, then run each file under browser-extension/tests/ with Node. See architecture, deployment, privacy, and production extraction map.
License
Common AI Memory is source-available under the PolyForm Noncommercial License 1.0.0.
Noncommercial use is allowed.
Noncommercial modification and redistribution are allowed under the license terms.
Commercial use is not permitted.
Redistributed or modified copies must preserve the required attribution notice:
Original project created by Ilyra.
See LICENSE for the controlling notice and the official PolyForm license URL.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections. Dates show when Glama detected each change.
No tool schema history has been recorded yet.
This server cannot be installed
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Related MCP Connectors
Persistent AI memory shared across Claude, ChatGPT, coding agents, and compatible MCP clients.
Shared long-term memory vault for AI agents with 20 MCP tools.
One memory, every AI. A shared, user-owned markdown memory your AI clients read and write over MCP.
One memory, every AI: Claude, ChatGPT, Perplexity, Gemini, Cursor, OpenClaw, Hermes, any MCP client.
Related MCP Servers
- AlicenseNot gradedqualityBmaintenanceEnables AI agents with persistent, multi-layer memory, multi-agent collaboration rooms, and video generation, all accessible via MCP tools.1MIT
- FlicenseNot gradedqualityDmaintenanceEnables multi-agent communication between AI agents via MCP tools with real-time message routing, admin control, and dual-language support.8-
- AlicenseNot gradedqualityBmaintenanceEnables AI agents to share knowledge, coordinate tasks, and maintain persistent memory across distributed infrastructure with secure vaults and 130+ MCP tools.7MIT
- AlicenseNot gradedqualityCmaintenanceMCP server that provides AI agents with persistent memory, cross-agent sharing, and context management, enabling them to remember conversations, track complex tasks, and evolve skills across tools.2MIT
Latest Blog Posts
- Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)By Om-Shree-0709 on .mcpAgent IdentityOAuth 2.1
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/IlyraVale/common-ai-memory'
If you have feedback or need assistance with the MCP directory API, please join our Discord server