What is AgentOS?

The runtime for your agent platform.

AgentOS is a FastAPI app that runs your agent platform. It's a live service that:

  1. Runs agents on demand, via API requests or interfaces like Slack, Telegram, and others.
  2. Maintains long-running sessions that last from minutes to days to weeks.
  3. Is durable across restarts, replicas, and infrastructure failures.
  4. Is secure against unauthenticated access.
  5. Logs every run and action taken.

Build your agents using the Kern SDK, run them using the AgentOS runtime.

Here's a minimal example of the AgentOS serving an Agent with memory, state and MCP:

1from kern.agent import Agent
2from kern.db.sqlite import SqliteDb
3from kern.models.anthropic import Claude
4from kern.os import AgentOS
5from kern.tools.mcp import MCPTools
6
7agno_assist = Agent(
8 name="Kern Assist",
9 model=Claude(id="claude-sonnet-4-5"),
10 db=SqliteDb(db_file="kern.db"),
11 tools=[MCPTools(url="https://kern.ndx.rocks/mcp")],
12 add_datetime_to_context=True,
13 add_history_to_context=True,
14 num_history_runs=3,
15 markdown=True,
16)
17
18agent_os = AgentOS(agents=[agno_assist])
19app = agent_os.get_app()

Key Features

  • Production API: 50+ ready to use endpoints with SSE-compatible streaming.
  • Data Ownership: Sessions, memory, knowledge, and traces stored in your database.
  • Request Isolation: No state bleed between users, agents, or sessions.
  • Security: JWT-based RBAC with hierarchical scopes.
  • Observability: Traces stored in your database with no third-party egress or vendor lock-in.
  • Governance: Guardrails, human-in-the-loop, and approval flows for full control.
  • Multi-framework: Serve agents built with the Claude Agent SDK, LangGraph and DSPy alongside native Kern agents. See Multi-Framework Support.

Architecture

AgentOS has two parts:

  1. Runtime: A FastAPI service that runs agents, teams, and workflows.
  2. Control plane: A UI for managing, monitoring, and debugging your system.

The runtime exposes the APIs that power both the control plane and your AI products. The runtime can serve agents built with the Kern SDK, the Claude Agent SDK, LangGraph, or DSPy.

AgentOS Architecture

The AgentOS runs as a container in your cloud. The UI connects directly from the browser. No proxies. No data relays. Your data stays completely private.

Private by Design

Most AI tooling stores your data on their servers. You pay retention costs, deal with egress fees, and depend on their security. AgentOS runs entirely in your infrastructure:

  • Your database: Sessions, memory, knowledge, traces. All stored where you control it.
  • Zero transmission: No conversations, logs, or metrics sent to Kern.
  • From browser to runtime: The control plane connects from your browser to the runtime. Kern stores no data except for your runtime endpoint. All data resides in your database.

See AgentOS Security for more details.

AgentOS Security and Privacy Architecture

Next Steps