Agent
Agent
PRIMARY AGENT
πŸ‡ΊπŸ‡Έ --:-- --
ENGINE
Engine State
β€”
β€”
⚑ β€”
cycle β€”
β€”
β€”
β€”
This Agent Surface
Home belongs to this dashboard agent.
🧠 Brain Storage Disk truth vs live memory. Mismatch = silent data loss. Should always match.
🩺 Live Problems Ground truth β€” auto-verified every ~90s, auto-remediated before escalation.
Seeded problems have seedOrigin: system β€” editing the spec is fine, they'll re-seed with defaults on engine restart.
Good Life UNKNOWN
Edit Problem
Verifier types: file_mtime, file_exists, pm2_status, http_ping, disk_free, graph_not_empty
Remediator types: pm2_restart (home23-* only), run_shortcut, fetch_url, exec_command (clean_pm2_logs, clean_npm_cache), dispatch_to_agent (hands the problem to the current agent with full tools, args.budgetHours), notify_jtr
⚑ Autonomous Actions
Settings β†’ Agency → curate the allow-list, toggle dry-run, view rejected requests
πŸ”” NOTIFY Stream (debug) drained by the promoter worker into live-problems
✨ Signals Wins, resolutions, positive observations β€” last 48h
COSMO: checking...
πŸ“₯ Ingestion Compiler
Tile Action

Worker Desk

Ask reusable specialists to inspect Home23, verify fixes, and hand useful findings back to the house agents.
Worker Library
β€”
β€”
β€”
β€”
Start From A Need
Installed Specialists
Proof Trail
Proof Receipt

Select a completed check to see what was inspected, what passed, what failed, and what Jerry can learn from it.

Loading brain map...

Intelligence

β€”Nodes
β€”Edges
β€”Cycles
β€”Compiled

Self-Understanding

Awaiting synthesis...

Consolidated Insights

Knowledge Index

Recent Activity

    Home23

    An installable AI operating system β€” persistent agents with living brains.

    What This Is

    Home23 is a complete AI operating system that runs on your machine. Your agent thinks autonomously through cognitive loops, grows a persistent brain over time, dreams during idle periods, and is reachable through this dashboard, Telegram, an AI IDE, and a research engine.

    The Four Systems

    🧠 Agent

    Always-on AI with cognitive loops, 26+ tools, and a brain that grows through conversation, document ingestion, and autonomous thought.

    πŸ”¬ cosmo23

    Autonomous research engine. Launch deep research runs that spawn agent swarms, explore the web, and build knowledge brains.

    ⚑ evobrew

    AI IDE. Query any brain, write code with AI, talk directly to your agents. The power tool for advanced work.

    🏠 Dashboard

    This screen. Chat with your agent, see what it's thinking, watch it dream, launch research, manage settings.

    How It Works

    The cognitive engine runs continuous think-consolidate-dream cycles. During waking hours, it processes thoughts, pursues goals, and responds to messages. During sleep periods, it dreams β€” synthesizing connections across its brain, consolidating knowledge, and growing.

    Documents fed through the feeder are LLM-synthesized before brain entry: raw text becomes structured knowledge with extracted concepts, relationships, and insights.

    Dashboard Guide

    Home β€” Live view of thoughts, vibe images, chat, brain log, dream log, feeder status
    Intelligence β€” Scheduled synthesis of brain knowledge, insights, and self-understanding
    Brain Map β€” 3D force-directed visualization of all brain nodes and connections
    Settings β€” Configure providers, agents, models, channels, and system behavior
    cosmo23 β€” Full research engine UI for launching and managing research runs
    evobrew β€” Opens the AI IDE in a new tab with your agent pre-selected

    Embedding Provider

    Your agent's brain uses vector embeddings for semantic memory. Pick one embedding model and stick with it β€” changing models later means re-embedding your entire brain. Ollama with nomic-embed-text is recommended (free, works well for continuous operation).

    Home23 Native Chat
    Loading agent…