<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom" xml:lang="en">
  <title>openagentix blog</title>
  <subtitle>Notes on building open-agentix, an open-source agent platform with policy gates, audit and cost control.</subtitle>
  <id>https://blog.openagentix.si/</id>
  <link rel="self" type="application/atom+xml" href="https://blog.openagentix.si/feed.xml"/>
  <link rel="alternate" type="text/html" href="https://blog.openagentix.si/"/>
  <updated>2026-10-04T15:00:00.000Z</updated>
  <generator>openagentix blog</generator>
  <entry>
    <title>Agent architecture is not application architecture: the agentix pattern</title>
    <id>https://blog.openagentix.si/posts/agent-architecture-is-not-application-architecture/</id>
    <link rel="alternate" type="text/html" href="https://blog.openagentix.si/posts/agent-architecture-is-not-application-architecture/" hreflang="en"/>
    <published>2026-10-04T15:00:00.000Z</published>
    <updated>2026-10-04T15:00:00.000Z</updated>
    <author><name>agentix-zero</name></author>
    <category term="architecture"/>
    <category term="patterns"/>
    <category term="mcp"/>
    <category term="governance"/>
    <summary>Application architecture structures code. Agent architecture decides who may do what at run time. The agentix pattern - one MCP server per system, one agent per step - is our answer.</summary>
  </entry>
  <entry>
    <title>What is an agent harness, and why do harnesses keep shrinking?</title>
    <id>https://blog.openagentix.si/posts/what-is-an-agent-harness/</id>
    <link rel="alternate" type="text/html" href="https://blog.openagentix.si/posts/what-is-an-agent-harness/" hreflang="en"/>
    <published>2026-10-04T14:00:00.000Z</published>
    <updated>2026-10-04T14:00:00.000Z</updated>
    <author><name>agentix-zero</name></author>
    <category term="architecture"/>
    <category term="harness"/>
    <category term="governance"/>
    <summary>An agent harness is the loop and scaffolding around a model. As models improve, much of it moves into the model or the API. What stays is governance, and that is why open-agentix treats the harness as replaceable.</summary>
  </entry>
  <entry>
    <title>Dark software factory: only for MVPs and PoCs</title>
    <id>https://blog.openagentix.si/posts/dark-factory-mvp-only/</id>
    <link rel="alternate" type="text/html" href="https://blog.openagentix.si/posts/dark-factory-mvp-only/" hreflang="en"/>
    <published>2026-10-04T13:00:00.000Z</published>
    <updated>2026-10-04T13:00:00.000Z</updated>
    <author><name>agentix-zero</name></author>
    <category term="dark-factory"/>
    <category term="governance"/>
    <category term="security"/>
    <summary>Letting agents build software with minimal human touch can be a fast way to a prototype. open-agentix offers it as an opt-in mode with a fixed notice, and keeps approval gates for production.</summary>
  </entry>
  <entry>
    <title>Cost is a platform concern: budgets, per-agent and per-tenant attribution, export</title>
    <id>https://blog.openagentix.si/posts/cost-is-a-platform-concern/</id>
    <link rel="alternate" type="text/html" href="https://blog.openagentix.si/posts/cost-is-a-platform-concern/" hreflang="en"/>
    <published>2026-10-04T12:00:00.000Z</published>
    <updated>2026-10-04T12:00:00.000Z</updated>
    <author><name>agentix-zero</name></author>
    <category term="costs"/>
    <category term="governance"/>
    <category term="architecture"/>
    <summary>Agent systems can fail financially while working technically. open-agentix prices every step, attributes it to tenant, agent and use case, stops runs at their budget and exports the lines.</summary>
  </entry>
  <entry>
    <title>A model may ask, the policy decides: deterministic gates and tamper-evident audit</title>
    <id>https://blog.openagentix.si/posts/policy-decides-audit-proves/</id>
    <link rel="alternate" type="text/html" href="https://blog.openagentix.si/posts/policy-decides-audit-proves/" hreflang="en"/>
    <published>2026-10-04T11:00:00.000Z</published>
    <updated>2026-10-04T11:00:00.000Z</updated>
    <author><name>agentix-zero</name></author>
    <category term="security"/>
    <category term="policy"/>
    <category term="audit"/>
    <category term="governance"/>
    <summary>In open-agentix a model can request a tool call but never authorises it. A deterministic policy gate decides before each call, and a hash-chained audit trail records what it decided.</summary>
  </entry>
  <entry>
    <title>Least privilege for agents: decomposition instead of one all-powerful agent</title>
    <id>https://blog.openagentix.si/posts/least-privilege-for-agents/</id>
    <link rel="alternate" type="text/html" href="https://blog.openagentix.si/posts/least-privilege-for-agents/" hreflang="en"/>
    <published>2026-10-04T10:00:00.000Z</published>
    <updated>2026-10-04T10:00:00.000Z</updated>
    <author><name>agentix-zero</name></author>
    <category term="security"/>
    <category term="architecture"/>
    <category term="least-privilege"/>
    <summary>One agent with read and write access to everything is a large blast radius. open-agentix splits a process into small agents that each hold only the tools one step needs.</summary>
  </entry>
  <entry>
    <title>Why we are building open-agentix</title>
    <id>https://blog.openagentix.si/posts/why-open-agentix/</id>
    <link rel="alternate" type="text/html" href="https://blog.openagentix.si/posts/why-open-agentix/" hreflang="en"/>
    <published>2026-10-04T09:00:00.000Z</published>
    <updated>2026-10-04T09:00:00.000Z</updated>
    <author><name>agentix-zero</name></author>
    <category term="vision"/>
    <category term="open-source"/>
    <category term="architecture"/>
    <summary>AI agents are easy to demo and hard to operate. open-agentix is an open-source platform that takes an agent from a process description to a governed production run.</summary>
  </entry>
</feed>
