Memory to Skill: Turn Agent Memory into Skills with MemSearch

🔗 GitHub: https://github.com/zilliztech/memsearch 🎬 Previous MemSearch video: Using Open-Source memsearch, Giving Your AI Agents Persistent, Human-Readable Memory: https://youtu.be/bf9eiFEs1JA Memory to Skill is a MemSearch feature that distills recurring workflows from persistent agent memory into reusable skill candidates. It learns from historical user-agent interactions, keeps the source journals behind every candidate, and gives users control over which skills are installed. Why Memory to Skill: ✦ Workflow distillation — turns repeated agent workflows into reusable skill candidates ✦ Outcome-driven learning — learns from completed work, not only repeated prompts ✦ Human-readable evidence — keeps the original Markdown journals available for review ✦ Continuous evolution — revises, improves, and consolidates candidates as new memory arrives ✦ Human-in-the-loop control — lets users inspect revision history before installation ✦ Flexible installation — supports project-level or global agent skill directories In this video, we show how to enable Memory to Skill in Codex, inspect the journals recorded by MemSearch, review two candidate skills mined from everyday agent use, install them in a project-level skill directory, restart Codex, and use a newly installed skill in a real workflow. ✨ If MemSearch is helpful to you, we would really appreciate a star on GitHub. We would also love to hear your thoughts, use cases, and feedback. Chapters: 00:00 Introduction to Memory to Skill 00:24 Why agent skills need a better creation workflow 01:32 From memory to reusable skills 02:02 How Memory to Skill works 03:42 Demo setup 04:28 How MemSearch records agent journals 05:11 Candidate skills mined from daily work 06:05 Review and install candidates 07:14 Use an installed skill in Codex 07:51 Recap Find us on: Zilliz Website: https://zilliz.com Milvus Website: https://milvus.io Zilliz LinkedIn: https://www.linkedin.com/company/zilliz/ Milvus LinkedIn: https://linkedin.com/company/67143129 Milvus X: https://x.com/milvusio Milvus GitHub: https://github.com/milvus-io/milvus Try Zilliz Cloud: https://cloud.zilliz.com/signup