Back to directory

Kronos Agent OS (KAOS)

A self-hosted runtime for long-running AI agents that keep memory across sessions, load skills on demand, call MCP tools, run scheduled automations, and optionally coordinate sub-agents.

Kronos Agent OS (KAOS)

Kronos Agent OS (KAOS) is a local agent loop reachable through a CLI as well as Telegram, Discord, webhook and cron entry points. Its memory layer combines session history, full-text search recall, vector memory, a knowledge graph and periodic consolidation, and its skills system lets the agent load workspace-local procedures and references on demand.

A tool gateway exposes MCP tools, custom tools and browser tools with audit-friendly execution, and scheduled automations can run digests, monitoring, analytics and competitor-monitoring jobs against sources such as Zabbix, Grafana, Sentry, PostHog, Supabase and app stores. An optional swarm mode coordinates sub-agents with SQLite-based arbitration, and a dashboard exposes memory, jobs, tool calls and system status.

By default dynamic tool loading, dynamic MCP server management and SSH or server operations are disabled unless explicitly enabled. It installs from PyPI, requires Python 3.11 or newer plus Node.js for an optional dashboard, and can run offline in a demo mode with no LLM key. It is MIT licensed.

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.