<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Oxelan Signal — Analysis</title>
    <link>https://blog.oxelan.com</link>
    <description>The mechanism and the production consequence. Why a change matters, who owns it, and what to do next.</description>
    <language>en-us</language>
    <lastBuildDate>Tue, 27 Oct 2026 00:00:00 GMT</lastBuildDate>
    <managingEditor>hello@oxelan.com (Kusal Tharindu)</managingEditor>
    <copyright>© 2026 Oxelan</copyright>
    <atom:link href="https://blog.oxelan.com/format/analysis/rss.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>MCP is turning agents into a network. Treat it like one.</title>
      <link>https://blog.oxelan.com/ai/mcp-agent-trust-boundary/</link>
      <guid isPermaLink="true">https://blog.oxelan.com/ai/mcp-agent-trust-boundary/</guid>
      <pubDate>Tue, 27 Oct 2026 00:00:00 GMT</pubDate>
      <description>Agents now call agents over MCP. Each hop forwards instructions the origin never saw, and tool registries keyed by name are a spoofing surface. The boundary has to move from the model to the protocol.</description>
      <category>ai-agents</category>
      <category>mcp</category>
      <category>llm-security</category>
      <category>prompt-injection</category>
    </item>
    <item>
      <title>The first 60 minutes of a supply-chain incident</title>
      <link>https://blog.oxelan.com/security/supply-chain-first-60-minutes/</link>
      <guid isPermaLink="true">https://blog.oxelan.com/security/supply-chain-first-60-minutes/</guid>
      <pubDate>Sat, 24 Oct 2026 00:00:00 GMT</pubDate>
      <description>A package compromise is a race between your rebuild and the attacker&apos;s next publish. Here is the order of operations that limits the blast radius.</description>
      <category>supply-chain</category>
      <category>incident-response</category>
      <category>ci-cd</category>
    </item>
    <item>
      <title>Open-weight models: the bill arrives after the benchmark</title>
      <link>https://blog.oxelan.com/ai/open-weight-models-economics/</link>
      <guid isPermaLink="true">https://blog.oxelan.com/ai/open-weight-models-economics/</guid>
      <pubDate>Fri, 23 Oct 2026 00:00:00 GMT</pubDate>
      <description>Open weights remove a license fee, not the operating cost. The decision hinges on utilization, latency, privacy and the team you can staff.</description>
      <category>open-weights</category>
      <category>inference</category>
    </item>
    <item>
      <title>Agentic AI changes the threat model, not just the prompt</title>
      <link>https://blog.oxelan.com/ai/agentic-threat-model/</link>
      <guid isPermaLink="true">https://blog.oxelan.com/ai/agentic-threat-model/</guid>
      <pubDate>Tue, 20 Oct 2026 00:00:00 GMT</pubDate>
      <description>Once a model can call tools, the interesting asset is no longer the prompt. It is the identity, the data in context, and the actions you forgot to scope.</description>
      <category>ai-agents</category>
      <category>llm-security</category>
      <category>threat-model</category>
    </item>
  </channel>
</rss>
