<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>小木同学 Builds</title><description>Full-stack products and backend systems — built by Leon</description><link>https://xiaomust.cn/</link><item><title>Not Sure What Agent Project to Build? Read These 7 Open-Source Repositories</title><link>https://xiaomust.cn/writing/seven-open-source-agent-projects/</link><guid isPermaLink="true">https://xiaomust.cn/writing/seven-open-source-agent-projects/</guid><description>Seven open-source Agent repositories worth studying: four for real product scenarios, then three for tool use, context management, and multi-agent design.</description><pubDate>Thu, 29 Oct 2026 12:00:00 GMT</pubDate><category>AI Agents</category><category>Open Source</category><category>Internship Projects</category></item><item><title>I Really Don&apos;t Recommend Another RAG Internship Project</title><link>https://xiaomust.cn/writing/stop-using-rag-as-your-internship-project/</link><guid isPermaLink="true">https://xiaomust.cn/writing/stop-using-rag-as-your-internship-project/</guid><description>RAG is still useful, but it should no longer be the same résumé centerpiece everyone builds. Agent tooling, planning, multi-step execution, and recovery show more differentiation.</description><pubDate>Tue, 20 Oct 2026 12:00:00 GMT</pubDate><category>RAG</category><category>AI Agents</category><category>Internship Projects</category></item><item><title>Overwhelmed by an Autumn AI Coding Task? Don&apos;t Start Coding Yet</title><link>https://xiaomust.cn/writing/autumn-ai-coding/</link><guid isPermaLink="true">https://xiaomust.cn/writing/autumn-ai-coding/</guid><description>A practical order for AI Coding tasks: make a checklist, run the starter project, verify the key paths, compare independent attempts, and submit the version you tested.</description><pubDate>Tue, 13 Oct 2026 12:00:00 GMT</pubDate><category>AI Coding</category><category>Campus Recruiting</category><category>Interviews</category></item><item><title>A useful agent should show its work</title><link>https://xiaomust.cn/writing/reviewable-agents/</link><guid isPermaLink="true">https://xiaomust.cn/writing/reviewable-agents/</guid><description>Why evidence, execution traces, and explicit failure boundaries matter more than a confident final answer.</description><pubDate>Tue, 29 Sep 2026 00:00:00 GMT</pubDate><category>AI Agents</category><category>Evidence</category><category>Architecture</category></item><item><title>The model proposes; the runtime decides</title><link>https://xiaomust.cn/writing/capability-gateways/</link><guid isPermaLink="true">https://xiaomust.cn/writing/capability-gateways/</guid><description>A capability gateway keeps tool use governed even when model behavior remains probabilistic.</description><pubDate>Sun, 20 Sep 2026 00:00:00 GMT</pubDate><category>Tool Calling</category><category>Guardrails</category><category>Backend</category></item><item><title>Design long-running agents for interruption</title><link>https://xiaomust.cn/writing/recoverable-agent-runs/</link><guid isPermaLink="true">https://xiaomust.cn/writing/recoverable-agent-runs/</guid><description>Checkpoints, durable events, and idempotent work turn disconnection from a disaster into a normal state transition.</description><pubDate>Fri, 11 Sep 2026 00:00:00 GMT</pubDate><category>Recovery</category><category>Async Jobs</category><category>SSE</category></item></channel></rss>