Qoopia
OfficialAllows connecting your own Telegram bot to chat with your Qoopia agent and access its memory.
Click on "Deploy 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., "@QoopiaRemember this project context for all my AI agents."
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.
Qoopia gives your agents persistent memory across sessions and clients. Use your own local installation or server, connect agents through MCP, and keep working in the tools you already use.
What you can do
Keep notes, files and session context accessible to your connected agents.
See agents, their notes and AgentComm conversations in a dashboard.
Choose Codex or Claude Code for My Qoopia agent, using your own eligible subscription. Chat in the dashboard or connect your own Telegram bot.
Connect separate environments through controlled bridges. Choose what to publish and what to import.
Turn approved work into reusable skill packages and share selected materials with other environments.
MCP access to memory and authorization to run a model are separate connections. Your agents do not gain access to somebody else's memory by registering on the website.
Related MCP server: Mycelia
Get started
Already use Codex or Claude Code with access to your computer? Give it this task:
Install Qoopia on this computer using https://qoopia.ai/docs#agent-install.
Read the instructions fully, verify the official download, and create my own
memory. Ask me to complete sign-ins and approve access. Connect my chosen
client and verify it with an actual memory call before reporting success.Or download the signed installer, follow the included instructions, and open Qoopia. No developer checkout is needed.
Platform | V1 support |
macOS | Apple Silicon, macOS 15 or later |
Linux | x64, glibc 2.34 or later; |
Windows / Intel Mac | Not included in this release |
ChatGPT Web/Desktop MCP connections are experimental. Managed HTTPS is a limited pilot; a complete real-account enrollment has not yet been qualified. A model subscription is subject to the provider's availability and limits. Telegram requires your own bot and a running, connected host.
Your data and control
SQLite is the canonical store in the installation you select. Access is scoped and revocable. Bridges do not merge everyone's private memory. External model calls and browser connections can send authorized data outside your machine; self-hosting does not mean every operation stays offline. Managed HTTPS uses Cloudflare Tunnel and is not end-to-end encrypted through that intermediary.
Explore the code
Area | Location |
Memory and MCP | |
Dashboard agent | |
Client connections | |
Bridges | |
Database evolution | |
Tests |
This is the public source distribution of Qoopia V1. It starts with a clean history and excludes private operational records. Source provenance records the original release revision and file checksums; published installers retain their existing signatures and release identifiers.
Develop
Install Bun 1.3.11, then:
git clone https://github.com/qoopia/qoopia-source.git
cd qoopia-source
bun install --frozen-lockfile
bun run typecheck
bun run lint
bun test
bun run test:storage-fullFor a disposable development installation, choose an unused directory:
export QOOPIA_ROOT="$(mktemp -d)"
export QOOPIA_SERVER_ROLE=canonical
bun run migrate
bun run devThe ordinary suite leaves one deliberate skip for the SQLite-full fault. test:storage-full runs it in a fresh temporary database; both commands are mandatory in CI. See the storage-full check.
Use synthetic data. Keep production credentials and personal memory out of tests. The schema-35 upgrade test includes a verified source-only fixture, so it does not need access to private Git history.
See contribution guidelines and security reporting.
License
MIT. Bundled third-party components retain their own license notices in scripts/vendor and the brand asset directories.
This server cannot be deployed
Maintenance
Related MCP Connectors
Shared, governed long-term memory for AI agents across tools and sessions via MCP and REST.
Persistent personal memory for AI assistants — save, search, and recall across every MCP client.
Persistent AI memory shared across Claude, ChatGPT, coding agents, and compatible MCP clients.
Persistent memory for AI agents across Claude, ChatGPT and any MCP client.
Related MCP Servers
- AlicenseAqualityDmaintenanceProvides persistent memory for AI assistants via MCP, enabling them to store and recall facts, preferences, and tasks across conversations using either local file storage or a cloud backend with semantic search.56 npmMIT
- AlicenseNot gradedqualityBmaintenanceProvides a persistent, cross-tool memory layer for AI coding agents via MCP, enabling storage and retrieval of decisions, preferences, and context across different tools and models.5 npm1MIT
- AlicenseNot gradedqualityCmaintenanceProvides persistent, cloud-based memory for AI agents, letting them remember preferences, skills, and decisions across sessions via an MCP-compatible interface.Apache 2.0
- AlicenseNot gradedqualityCmaintenanceProvides persistent, multi-project memory for AI agents via MCP, enabling them to store and retrieve tasks, decisions, errors, checkpoints, and documentation while keeping context isolated by project.2MIT