CORE: memory and tasks for Claude Code
CORE
The CORE plugin for Claude Code keeps decisions and project context in a knowledge graph and gives them to the agent each session
- 2,003 ★ on the repository
- Last commit
Broad access
Why this level
- Hooks send records of Claude Code sessions to app.getcore.me and use your account token
- The platform reads email, calendar, trackers and chats, and the agents it starts run commands on the machine
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/coreDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Hooks send records of Claude Code sessions to app.getcore.me and use your account token
- The platform reads email, calendar, trackers and chats, and the agents it starts run commands on the machine
This is third-party code. Review the repository files before installing.
What it does
The core_brain plugin connects Claude Code to CORE memory. It installs from the redplanethq marketplace and adds the remote MCP server app.getcore.me and two session hooks. Signing in needs the corebrain command-line client. CORE itself is an open-source personal AI assistant you can deploy on your own server. It builds a temporal knowledge graph from email, meetings, work services such as GitHub and Slack, and conversations with assistants. Each task gets a spec, a plan and its own chat thread. There is a daily notes page called the Scratchpad, voice input and access from messengers. CORE also starts Claude Code or Codex sessions with the full task context on your machine or in Docker and opens a pull request when they finish.
Who it is for. For developers who want Claude Code to remember decisions and context between sessions and are ready to keep that memory in the external CORE service.
Good fit when
- Give Claude Code memory of past decisions, preferences and project context without re-explaining
- Turn a line on the daily notes page into a task: CORE gathers context from the repo and memory and proposes a plan
- Run long Claude Code or Codex sessions that continue with the laptop lid closed
- Send tasks to the assistant from WhatsApp, Slack or Telegram and get the result into the same context
- Connect email, calendar, Linear, Sentry and Notion to the agent through one set of MCP tools
Not a fit when
- You need simple local memory in files, with no server and no account
- Session data must not leave for an external server
- You only need a skill or a short command set for Claude Code: CORE is built as a whole platform with connectors and tasks
Example request
Recall what we decided about authorization in the last session and continue the task from there.Limitations
The plugin connects to the app.getcore.me cloud service and needs an account there. The CORE platform itself can run on your own server, which needs Docker 20.10, Docker Compose 2.20, 4 CPU cores and 8 GB of memory.
The license is AGPL-3.0 with additional terms. Read the LICENSE file before using it in a company.
Voice hotkeys, the Mac app and some connectors are not part of the plugin; their setup is in the documentation at docs.getcore.me. It needs Claude or Codex subscriptions or provider keys, and their availability from Russia should be checked separately.
How to disable. Uninstall the plugin with `/plugin uninstall core_brain@redplanethq` and the marketplace with `/plugin marketplace remove redplanethq`. Sign out through the corebrain client, and delete data already stored in CORE inside the service.
FAQ
How do I connect CORE memory to Claude Code?
Install the client with `npm install -g @redplanethq/corebrain` and sign in with `corebrain login`. Then in Claude Code run `/plugin marketplace add redplanethq/core` and `/plugin install core_brain@redplanethq`.
What do the core_brain plugin hooks do?
The SessionStart hook fires on startup, clear and compact and calls the CORE API. The Stop hook fires when a session stops and sends records to CORE. Each hook has a 300-second timeout.
How do I run CORE on my own server?
Run `npm install -g @redplanethq/corebrain && corebrain setup`. The wizard asks for a directory, AI provider and key, generates secrets, starts the stack and opens http://localhost:3033. A Railway deployment is available for servers.
How is CORE different from memory plugins such as claude-mem?
Memory plugins store observations from agent sessions. CORE is broader: a platform with a knowledge graph built from email, meetings and trackers, a task queue, agent runs and messenger access.
Which apps can I connect to CORE?
More than fifty apps connect through MCP: GitHub, Linear, Jira, Slack, Gmail, Calendar, Sentry, Notion and Todoist. Webhooks turn inbound events into tasks.
Discussion
No questions or reviews yet.
Related
PluginsOpen-source personal AI assistant on your own machine: answers in Telegram, Slack, Discord and WhatsApp, extended with skills and plugins
An engineering system for coding agents: 68 agents, 292 skills, hooks, memory, rules and the AgentShield security scanner
DeepSeek Harness is an open-source AI agent from DeepSeek where every feature is a plugin. It runs in the browser, on macOS and on Windows.
A self-improving agent from Nous Research with a TUI, messaging gateway, cron jobs and skills it writes itself
An open source coding agent for the terminal and desktop with build and plan modes
Ollama runs open models on your own machine and serves them to agents through a local API compatible with OpenAI and Anthropic.
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Repository stars | Main difference |
|---|---|---|---|
| CORE: memory and tasks for Claude Code (this entry) | no | 2,003 | |
| Claude-Mem | no | 102,612 | a persistent memory plugin built from session observations, with no task queue, connectors or agent runs |
| claude-supermemory | yes | 2,794 | a plugin for the Supermemory service with prompt-driven search and a codebase index, with no personal task platform |
| base (basemode) | no | 306 | a Rust CLI that builds a knowledge graph of code and decisions and |
README in short
The project calls CORE a personal AI operating system that runs all the time and hands tasks to coding agents. On the LoCoMo memory benchmark CORE reaches 88.24% average accuracy.