<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Derik Callahan — Writing</title><description>Notes on AI risk, agentic security, and the projects I build.</description><link>https://derikcallahan.com/</link><item><title>A resume for two audiences</title><link>https://derikcallahan.com/writing/a-resume-for-two-audiences/</link><guid isPermaLink="true">https://derikcallahan.com/writing/a-resume-for-two-audiences/</guid><description>More and more, the first reader of a resume is an AI agent. So I built this site to be read by both people and agents — Markdown twins, llms.txt, JSON Resume, content negotiation, and a small read-only MCP server.</description><pubDate>Fri, 16 Oct 2026 00:00:00 GMT</pubDate><category>this site</category><category>MCP</category><category>llms.txt</category><category>web security</category></item><item><title>The agent is an identity</title><link>https://derikcallahan.com/writing/the-agent-is-an-identity/</link><guid isPermaLink="true">https://derikcallahan.com/writing/the-agent-is-an-identity/</guid><description>Most AI risk conversations start with the model. I think they should start with the credentials. An agent can only do what its identity allows — so govern the identity.</description><pubDate>Wed, 07 Oct 2026 00:00:00 GMT</pubDate><category>AI risk</category><category>non-human identity</category><category>identity governance</category><category>Zero Trust</category></item><item><title>Agent traces are security telemetry</title><link>https://derikcallahan.com/writing/agent-traces-are-security-telemetry/</link><guid isPermaLink="true">https://derikcallahan.com/writing/agent-traces-are-security-telemetry/</guid><description>If you can&apos;t replay what an agent saw, decided, and did, you can&apos;t investigate it. OpenTelemetry already gives us the format. The SOC needs to treat it like EDR data.</description><pubDate>Wed, 30 Sep 2026 00:00:00 GMT</pubDate><category>AI risk</category><category>OpenTelemetry</category><category>detection engineering</category><category>SOC</category></item><item><title>An agent for apps without APIs</title><link>https://derikcallahan.com/writing/an-agent-for-apps-without-apis/</link><guid isPermaLink="true">https://derikcallahan.com/writing/an-agent-for-apps-without-apis/</guid><description>Every identity program has a list of disconnected applications that someone provisions by hand. I built a proof of concept where a model drives the browser instead — with a human approving every write and a hashed screenshot behind every step.</description><pubDate>Sat, 12 Sep 2026 00:00:00 GMT</pubDate><category>identity governance</category><category>browser agents</category><category>human-in-the-loop</category><category>audit</category></item><item><title>Three gates for an offensive agent</title><link>https://derikcallahan.com/writing/three-gates-for-an-offensive-agent/</link><guid isPermaLink="true">https://derikcallahan.com/writing/three-gates-for-an-offensive-agent/</guid><description>I built an MCP server for HackerOne and a bug bounty pipeline around it. The interesting part isn&apos;t the 174 tools. It&apos;s the three places where the agent has to stop and wait for me.</description><pubDate>Sat, 22 Aug 2026 00:00:00 GMT</pubDate><category>agentic security</category><category>MCP</category><category>bug bounty</category><category>approval gates</category></item></channel></rss>