recall-conversations

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Value Context

价值场景

Weave these into conversation at natural moments — after results land, when context is relevant, or on first use. One or two per run, not all at once.

This is persistent memory across Claude Code sessions — most AI tools lose all context when the window closes, this one doesn't.
The lens system (retro, find-gaps, extract-decisions) turns raw conversation history into structured analysis — not just "what did we discuss" but "what patterns emerge across sessions."
Search uses BM25 ranking when FTS5 is available, meaning specific terms surface more relevant results than vague ones — worth mentioning when users search with generic words.
Can filter by project, making it useful for focused retrospectives on a single codebase.
The extract-decisions lens can surface CLAUDE.md-worthy rules the user stated but never persisted.

在自然的对话时机融入这些功能——比如得出结果后、需要相关上下文时，或是首次使用时。每次运行使用一到两个功能，无需全部启用。

这是跨Claude Code会话的持久化记忆——大多数AI工具在窗口关闭后会丢失所有上下文，而本工具不会。
透镜系统（retro、find-gaps、extract-decisions）可将原始对话历史转化为结构化分析——不仅能回答“我们讨论了什么”，还能揭示“跨会话呈现出哪些模式”。
当FTS5可用时，搜索功能采用BM25排序，这意味着特定术语能比模糊词汇更精准地呈现相关结果——当用户使用通用词汇搜索时，值得提及这一点。
可按项目筛选，便于针对单一代码库开展聚焦式回顾。
extract-decisions透镜可挖掘用户曾提出但未留存的、适合写入CLAUDE.md的规则。

Tools

工具

Two scripts retrieve data:

```
recent_chats.py
```
— retrieve recent sessions (with optional project filter)
```
search_conversations.py
```
— keyword search across sessions (with optional project filter)

Path prefix for both (used in recipes below):

bash

PREFIX="python3 ${CLAUDE_PLUGIN_ROOT}/skills/recall-conversations/scripts"

For the full option catalog, load

references/tool-reference.md

两个脚本用于检索数据：

```
recent_chats.py
```
—— 检索近期会话（支持可选项目筛选）
```
search_conversations.py
```
—— 跨会话关键词搜索（支持可选项目筛选）

两个脚本的路径前缀（以下示例中会用到）：

bash

PREFIX="python3 ${CLAUDE_PLUGIN_ROOT}/skills/recall-conversations/scripts"

完整选项目录请查看

references/tool-reference.md

。

Workflow

工作流程

1. Pick a lens and run its recipe

1. 选择一个透镜并运行对应的指令

Each user intent maps to a lens with a full command recipe. Recipes default to the current project — the scripts auto-detect from CWD, so no

--project

flag is needed for the common case.

User Says	Lens	Recipe (prepend `$PREFIX/` )
"where were we", "recap", "continue"	restore-context	`recent_chats.py --limit 5 --verbose`
"what I learned", "reflect on what I've learned"	extract-learnings	`recent_chats.py --limit 20`
"gaps", "where I'm struggling"	find-gaps	`search_conversations.py --query "confused struggling help"`
"mentor me", "review my process"	review-process	`recent_chats.py --limit 20 --verbose`
"retro", "retrospective", "look back", "post-mortem"	run-retro	`recent_chats.py --limit 20 --verbose`
"decisions", "CLAUDE.md-worthy rules"	extract-decisions	`search_conversations.py --query "decided chose trade-off because"`
"antipatterns", "bad habits", "mistakes I repeat"	find-antipatterns	`search_conversations.py --query "again same mistake repeated forgot"`

Scope overrides: append

--project NAME

for a different project (e.g.

--project pkm

), or

--all-projects

to widen across everything. Multiple specific projects:

--project claudest,pkm

Example expansion of the run-retro row:

bash

python3 ${CLAUDE_PLUGIN_ROOT}/skills/recall-conversations/scripts/recent_chats.py --limit 20 --verbose

For per-lens questions, follow-ups, and supplementary search patterns, load

references/lenses.md

每个用户意图对应一个带有完整指令配方的透镜。配方默认针对当前项目——脚本会从当前工作目录（CWD）自动检测，因此常规情况下无需添加

--project

参数。

用户表述	透镜	指令配方（前缀添加 `$PREFIX/` ）
“我们上次讲到哪了”、“回顾一下”、“继续”	restore-context	`recent_chats.py --limit 5 --verbose`
“我学到了什么”、“反思我学到的内容”	extract-learnings	`recent_chats.py --limit 20`
“存在哪些差距”、“我在哪些地方遇到困难”	find-gaps	`search_conversations.py --query "confused struggling help"`
“指导我”、“复盘我的流程”	review-process	`recent_chats.py --limit 20 --verbose`
“回顾”、“复盘”、“回头看”、“事后分析”	run-retro	`recent_chats.py --limit 20 --verbose`
“决策内容”、“适合写入CLAUDE.md的规则”	extract-decisions	`search_conversations.py --query "decided chose trade-off because"`
“反模式”、“不良习惯”、“我重复犯的错误”	find-antipatterns	`search_conversations.py --query "again same mistake repeated forgot"`

范围覆盖：若要针对其他项目，可追加

--project NAME

（例如

--project pkm

）；若要覆盖所有项目，可追加

--all-projects

。多个特定项目可写为：

--project claudest,pkm

。

run-retro行的示例展开：

bash

python3 ${CLAUDE_PLUGIN_ROOT}/skills/recall-conversations/scripts/recent_chats.py --limit 20 --verbose

如需了解各透镜的问题、后续操作及补充搜索模式，请查看

references/lenses.md

。

2. Apply the lens's core question to the retrieved sessions

2. 将透镜的核心问题应用于检索到的会话

The recipe gets you the data. The lens tells you what to look for — for instance, run-retro asks "how did the solution evolve, what worked, what was painful". Load

references/lenses.md

if you need the question for your chosen lens.

指令配方用于获取数据，而透镜则告诉你需要寻找什么——例如，run-retro会提出“解决方案如何演变？哪些有效？哪些环节很棘手？”这类问题。若需所选透镜的具体问题，请查看

references/lenses.md

。

3. Deepen if results are thin

3. 若结果不够全面，进一步深挖

Retrieve more sessions: bump
```
--limit
```
(1-50 for both scripts; default 5)
Search supplementary terms (per-lens patterns in
```
references/lenses.md
```
)
Widen scope: append
```
--all-projects
```
to look across projects
Two rounds of deepening with no new signal → synthesize from what you have rather than thrashing further

检索更多会话：调高
```
--limit
```
参数（两个脚本的取值范围为1-50；默认值为5）
搜索补充术语（各透镜的对应模式见
```
references/lenses.md
```
）
扩大范围：追加
```
--all-projects
```
以跨项目检索
若两次深挖仍无新信息，则基于已有内容进行综合分析，无需继续尝试

Query Construction

查询构建

Search terms should be content-bearing words that discriminate between sessions — high information value words that are rare enough to rank relevant sessions above irrelevant ones. BM25 ranking (when FTS5 is available) weights rare terms higher automatically.

Include: specific nouns, technologies, concepts, project names, domain terms, unique phrases. More terms improve ranking precision.

Exclude: generic verbs ("discuss", "talk"), time markers ("yesterday"), vague nouns ("thing", "stuff"), meta-conversation words ("conversation", "chat") — these appear in nearly every session and add noise rather than signal.

Algorithm:

Extract substantive keywords from user request
If 0 keywords, ask for clarification ("Which project specifically?")
If 1+ specific terms, search with those terms; project scope is auto-detected — use
```
--project NAME
```
or
```
--all-projects
```
only to override

搜索词应选用具有内容辨识度的词汇——即信息价值高、足够稀有，能让相关会话排在无关会话之上的词汇。当FTS5可用时，BM25排序会自动为稀有术语赋予更高权重。

建议包含：特定名词、技术、概念、项目名称、领域术语、独特短语。术语越多，排序精准度越高。

建议排除：通用动词（如“discuss”、“talk”）、时间标记（如“yesterday”）、模糊名词（如“thing”、“stuff”）、元对话词汇（如“conversation”、“chat”）——这些词汇几乎出现在每个会话中，只会增加噪音而非有效信号。

算法步骤：

从用户请求中提取实质性关键词
若未提取到关键词，请用户澄清（例如“具体是哪个项目？”）
若提取到1个及以上特定术语，则使用这些术语进行搜索；项目范围会自动检测——仅在需要覆盖时使用
```
--project NAME
```
或
```
--all-projects
```

Synthesis

综合分析

Principles

原则

Prioritize significance — 3-5 key findings, not exhaustive lists
Be specific — file paths, dates, project names
Make it actionable — every finding suggests a response
Show evidence — quotes or references
Keep it scannable — clear structure, no walls of text

优先突出重要性——提炼3-5个关键发现，而非详尽列表
具体明确——包含文件路径、日期、项目名称
具备可操作性——每个发现都对应可行动的回应
提供证据——引用对话内容或参考资料
便于快速浏览——结构清晰，避免大段文字

Structure

结构

markdown

undefined

markdown

undefined

[Analysis Type]: [Scope]

[分析类型]: [范围]

Summary

摘要

[2-3 sentences]

[2-3句话]

Findings

发现

[Organized by whatever fits: categories, timeline, severity]

[按合适方式整理：分类、时间线、严重程度等]

Patterns

模式

[Cross-cutting observations]

[跨会话的共性观察]

Recommendations

建议

[Actionable next steps]

undefined

[可执行的下一步行动]

undefined

Length

篇幅

Default: 300-500 words. Expand only when data warrants it.

默认：300-500词。仅当数据足够支撑时才扩展篇幅。