AgentOS: A Deterministic Event-Sourced Runtime for Auditable Agent Systems
By
david927
Toasted golden, schmeared with insight. Top of the rack.
Summary
The article describes AgentOS, an experimental agent substrate/sandbox system featuring a deterministic, event-sourced runtime where agent worlds are replayable from logs. Heavy logic runs in sandboxed WASM modules, and all real-world side effects (HTTP, LLM calls, code compilations) are capability-gated and recorded as signed receipts. The system enables auditable, reversible, and composable upgrades and automations, with a typed control-plane intermediate representation (AIR) that allows the system to treat its own schemas, modules, plans, and policies as data, evolving through a governed loop (propose > shadow-run > approve > apply).
Key quotes
· 4 pulledIt's a deterministic, event-sourced runtime where an 'agent world' is replayable from its log, heavy logic runs in sandboxed WASM modules, and every real-world side effect (HTTP, LLM calls, code compilations, etc.) is explicitly capability-gated and recorded as signed receipts.
It ensures that upgrades and automations are auditable, reversible, and composable.
The fun bit is a small typed control-plane intermediate representation (AIR) that lets the system treat its own schemas/modules/plans/policies as data and evolve via a governed loop (propose > shadow-run > approve > apply).
kind of 'Lisp machine vibes' but aimed at agents that need reliable self-modification rather than ambient scripts.
You might also wanna read
Agentspan: Open-source runtime for durable AI agent workflows with crash recovery and observability
Agentspan is an open-source server and SDK (MIT licensed) that enables developers to run AI agents as durable workflows. It provides crash r
Product Hunt·16d ago
Agent Sandbox: A Tool for AI Agents to Run Code and Generate Files Locally
Agent Sandbox is a tool that provides AI agents with sandboxed computing capabilities, allowing them to run Python/Bash scripts, install pac
Product Hunt·3mo ago
Runtime launches sandboxed coding agents for team collaboration with security guardrails
Runtime is a platform that turns coding agents into sandboxed, secure teammates that can be used across Slack, Linear, CLI, API, or browser.
Product Hunt·12d ago
OpenAI Updates Agents SDK with Codex-Style Harness and Enhanced Sandboxing
OpenAI's Build Hour session, led by engineer Steve Corley, introduced key updates to the Agents SDK, including a new "Codex-style harness" t
startuphub.ai·3d ago
AgentRail: Open-source local control plane for managing AI coding agent task lifecycles
AgentRail is an open-source, locally-run task lifecycle layer for AI coding agents. It solves the operational challenges beyond code writing
Product Hunt·17d ago
Secure AI Agent Deployment: Sandboxed Execution with relaxAI
This article promotes a webinar/presentation by Ben Norris, AI Engineer at relaxAI, focused on deploying AI agents within secure, sandboxed
civo.com·4d ago