Memory
Claude Code 新手安心神技 — coding · code-polish · memory
npx -y skills add JurnLoryu/claude-code-starter-kit --skill memoryAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
The user's persistent file-based memory + daily-log system, encoded as a skill. Sediments non-obvious learnings — root causes, established standards, pitfalls, user preferences — into memory files with an index, and auto-creates/updates a daily log of work done. Use whenever finishing a chunk of work, after a non-obvious fix or decision, when the user says to remember something, or to write the day's log. Triggers: remember this, sediment, write log, daily log, recap, 记一下, 记住, 沉淀, 写日记, 复盘, 总结今天.
SKILL.md
2.4 KB, as published. Nobody here has run it
Memory
把非显然的经验沉淀成带索引的记忆文件，并按「月/日」自动写工作日记——让 AI 有"记性"。
把"非显然的东西只踩一次坑"做成习惯：边干边沉淀记忆 + 自动写日记。
文件化记忆（~/.claude/projects/<project>/memory/）
- 一事一文件，frontmatter：
name(kebab，与文件名一致) /description(一句话，供召回判定) /metadata.type(user|feedback|project|reference)。 - 正文写事实；
feedback/project类补 Why 和 How to apply 两行。用[[name]]互链相关记忆（链到还没写的也行，等于标记待写）。 - 在
MEMORY.md加一行索引（标题 + 钩子）；MEMORY.md 只放索引，正文永远在各自文件。 - 去重而非堆叠：先找已覆盖同一事的文件直接改它，别新建重复；写错就删。
- 相对日期转绝对（"昨天"→具体日期）。
该记 / 不该记
- 记：非显然的修复根因、确立的标准与约定、踩过的坑、用户偏好与反馈。
- 不记：代码本身/git 历史/CLAUDE.md 已说明的；只对当前对话有用的。被要求记这类时，问"哪里非显然"，只记那个点。
自动日记（做完一段工作就写，不必用户开口）
- 位置沿用项目日志库的现有命名：
<X月日记>/<YY.M.D>/<YY.M.D>.md（例：记忆代码/六月日记/26.6.7/26.6.7.md）。 - 今天没有该文件就创建，已有就追加一节；用今天的真实日期。
- 一节写：做了什么 · 关键决策与理由 · 踩的坑与解法 · 改了哪些文件。
- 基于真实发生的事；没干事的休息日不补建空日志。
边界
记忆/日志库 ≠ 代码工程源。真实代码在各自仓库；记忆库只存记忆与日志，别当代码目录用。
Gives 0 of the 12 instructions most memory context skills give
Counted across 674 of the 847 authors here whose files we hold, read 2026-08-06
- inform the user when setup is completein 21 of 674, across 6 files
- confirm the draft with the user before writingin 21 of 674, across 6 files
- update the agent skills block in place if it existsin 21 of 674, across 6 files
- present findings to the userin 20 of 674, across 5 files
- write the three docs files from seed templatesin 20 of 674, across 5 files
- ask the user about each decision one at a timein 19 of 674, across 4 files
- edit CLAUDE.md if it existsin 18 of 674, across 3 files
- explore current repo statein 18 of 674, across 3 files
- do not overwrite user edits to surrounding sectionsin 18 of 674, across 3 files
- back up the original file before overwritingin 16 of 674, across 8 files
- keep the memory index under 200 linesin 15 of 674
- Provide actionable steps and verificationin 13 of 674, across 2 files
Said here and by no other author read
- include name, description, and type frontmatter
- add why and how to apply lines for feedback or project types
- cross-link related memories using name
- keep memory index file free of body text
- deduplicate existing memory instead of stacking duplicates
- delete memory written incorrectly
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once.