<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Holmes Lab</title><link>https://holmes-lab.web.app/en/</link><description>Engineering fieldnotes</description><item><title>Holmes-Kit</title><link>https://holmes-lab.web.app/en/projects/#holmes-kit</link><guid>https://holmes-lab.web.app/en/holmes-kit</guid><description>From requirements to code impact, connected by evidence.</description></item><item><title>Jarvis</title><link>https://holmes-lab.web.app/en/projects/#jarvis</link><guid>https://holmes-lab.web.app/en/jarvis</guid><description>Useful everyday AI — and the conditions for trusting it.</description></item><item><title>Holmes Lab</title><link>https://holmes-lab.web.app/en/projects/#hub</link><guid>https://holmes-lab.web.app/en/hub</guid><description>A place to understand the work and start a thoughtful conversation.</description></item><item><title>Start with intent</title><link>https://holmes-lab.web.app/en/methods/#spec-first</link><guid>https://holmes-lab.web.app/en/spec-first</guid><description>Good delegation begins with observable success criteria.</description></item><item><title>Show what works</title><link>https://holmes-lab.web.app/en/methods/#evidence</link><guid>https://holmes-lab.web.app/en/evidence</guid><description>Distinguish code, configuration, and observed behavior.</description></item><item><title>Design smaller permissions</title><link>https://holmes-lab.web.app/en/methods/#boundaries</link><guid>https://holmes-lab.web.app/en/boundaries</guid><description>Limit authority before refining the prompt.</description></item><item><title>Holmes-Kit: connect intent to code</title><link>https://holmes-lab.web.app/en/tools/holmes-kit/</link><guid>https://holmes-lab.web.app/en/tools/holmes-kit/</guid><description>For developers delegating implementation to an AI CLI who need requirements, design, tests and code connected with evidence for each change.</description></item><item><title>Agent Bridge: explicit collaboration across CLI sessions</title><link>https://holmes-lab.web.app/en/tools/agent-bridge/</link><guid>https://holmes-lab.web.app/en/tools/agent-bridge/</guid><description>Local collaboration between enrolled Codex/Claude sessions and managed AGY processes, separating delivery from acknowledgement. An alpha for early evaluators.</description></item><item><title>What should an AI check before changing code?</title><link>https://holmes-lab.web.app/en/notes/spec-first-ai-development/</link><guid>https://holmes-lab.web.app/en/notes/spec-first-ai-development/</guid><description>Add whitespace handling and a default name to one greeting function. Follow the same three tests from failure to success, connecting requirements, design decisions and execution evidence.</description></item><item><title>How can different AI CLI sessions collaborate explicitly?</title><link>https://holmes-lab.web.app/en/notes/cli-agent-handoff/</link><guid>https://holmes-lab.web.app/en/notes/cli-agent-handoff/</guid><description>A review request sent to another CLI has no reply. Should you send it again? Follow one illustrative request to separate delivery, acknowledgement, work completion and decisions under uncertainty.</description></item></channel></rss>