Back to directory

CORE

A self-hosted, always-on personal AI that indexes email, meetings, GitHub, Linear and Slack into a memory graph, then runs tasks through voice, a scratchpad, chat or messaging apps, spinning up coding-agent sessions when needed.

CORE

CORE is a self-hosted personal AI agent that indexes email, meetings, GitHub, Linear, Slack and prior assistant conversations into a temporal knowledge graph, so each task starts with existing preferences, decisions and context already loaded. Every unit of work is a task with a spec, a plan, live state and its own chat thread, run once or on a recurring schedule.

Tasks can be created from a daily scratchpad page, by voice through a keyboard shortcut, or from WhatsApp, Slack or Telegram, and CORE keeps the same memory and context regardless of which channel a task came from. For coding work, it spins up Claude Code or Codex sessions with full task context, runs them locally, in Docker or on Railway, and opens a pull request when finished, continuing to run after the user's laptop is closed. A built-in, isolated browser lets it act on the user's behalf without exposing the user's main browser session, and one-click integrations connect it to more than 50 apps over MCP, including GitHub, Linear, Jira, Slack, Gmail, Calendar, Sentry, Notion and Todoist.

It is open source and self-hosted.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.