Token导航 LogoToken导航TokenDH.com
运维权限需确认clawhub未标认证来源可访问clear审计通过

conversation-recap-to-obsidianconversation recap TO Obsidian 运维

Agent Skill

conversation-recap-to-obsidian 用于辅助前端页面、组件、样式和交互逻辑开发,适合在 OpenClaw 中需要维护前端项目、生成组件或检查界面实现时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

6,472

周安装

278

GitHub Stars

公开资料未说明

下载量

2,268
OpenClaw

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

MIT-0

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:conversation-recap-to-obsidian(conversation recap TO Obsidian 运维)
来源仓库:https://github.com/amortalsodyssey/conversation-recap-to-obsidian
安装命令:
openclaw skills install conversation-recap-to-obsidian
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 OpenClaw 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

ClawHubOpenClaw
openclaw skills install conversation-recap-to-obsidian

简介

conversation-recap-to-obsidian 将对话转化为 Obsidian 回顾笔记。

  • 适用于个人知识管理、每日/周总结生成场景。
  • 支持从 Markdown 或对话输入构建高价值笔记。
  • 使用前需配置 Obsidian 路径和模板。
  • 建议自定义标签体系以增强检索效率。conversation-recap-to-obsidian 属于运维类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

name
conversation-recap-to-obsidian
description
Build high-value Obsidian daily and weekly review notes from conversation or existing markdown notes. Use this skill whenever the user asks to summarize the current chat into Obsidian, append a structured session recap, regenerate a daily summary from a full daily note, create or refresh a weekly report, merge same-topic work across multiple days, group work by project/task instead of by date, or extract structured review notes with conclusions, key points, tags, and wikilinks.

Conversation Recap to Obsidian

This skill turns raw conversation or existing Obsidian markdown into review-ready notes, not generic summaries. Treat the whole daily/weekly note as input content regardless of who wrote each part.

The default goal is to help the user answer:

  • What were the main things done?
  • What problems did those things solve?
  • What were the key points?
  • What conclusions or outputs matter later?
  • Which notes/documents should be linked for follow-up?

Use this skill for

  • writing a structured recap of the current conversation into Obsidian
  • appending a single session entry into a daily note
  • regenerating a daily summary by reading the full daily note first
  • creating a weekly report from multiple daily notes
  • merging a multi-day thread into one weekly module
  • replacing stale generated summary blocks while preserving all non-target content
  • producing Obsidian wikilinks for relevant artifacts
  • consolidating notes that may contain mixed content from humans, this assistant, and other AI/tools

Core design

Split the work in three layers:

  1. Entry layer = raw work-item capture

- append a single session recap into the daily note - preserve concrete issue / solution / conclusion / key point details - act as the source material for later daily and weekly synthesis

  1. Daily summary layer = same-day aggregation

- read the full daily note - merge duplicate work threads - compress the day into a small number of reusable conclusions

  1. Weekly layer = cross-day synthesis

- merge same-topic work across multiple days - rank larger / more complex items first - avoid day-by-day流水账

Use scripts when they improve reliability. Do not avoid them just to stay “pure prompt only.”

Daily note target

Default path:

  • daily/YYYY/MM/YYYY-MM-DD.md (organized by year and month)
  • Users may override the base directory through local config

Mode 1: Session recap mode

Use this when the user has just finished one conversation or one work block and wants to record a new entry instead of refreshing the whole day.

Session recap principle

Create a new item in the daily note as source material for later summaries.

Default entry structure

#### 事项标题 — HH:mm

- **问题**: ...
- **方案**: ...
- **结论**: ...
- **关键点**: ...
- **关联**: [[...]] · [[...]]
- **标签**: #tag-a #tag-b

Session recap guidance

  • Prefer one concrete work item per entry.
  • If the conversation truly covered multiple unrelated things, either split into 2 entries or name the entry at a higher level.
  • Keep each field tight and useful.
  • The title should describe the work item, not just say “对话总结”.
  • Use tags sparingly; 1-3 strong tags are enough.
  • Preserve document hierarchy: session entries belong in the raw entry section of the daily note, and the generated ## 今日总结 block should stay as a higher-level summary section near the end of the note.
  • When appending an entry to a daily note that already contains ## 今日总结, insert the new entry before the generated summary block, then refresh the summary if needed.

Mode 2: Daily summary mode

A daily summary is regenerable. It is not an append-only log.

When asked to refresh the daily summary:

  1. Read the full daily note.
  2. Treat the whole note as usable source material regardless of whether parts were written by a person or another AI/tool.
  3. Ignore only the previous generated summary block for this skill to avoid recursive self-copying.
  4. Extract the day’s main work items from the note content.
  5. Compress them into a concise review section.
  6. Replace only the generated summary block.

Default daily output shape

## 今日总结

- 今日主要事项:...
- 核心解决的问题:...
- 关键点:...
- 结论/产出:...
- 相关文档:[[...]] · [[...]]
- 标签:#tag-a #tag-b

Daily writing guidance

  • Prefer outcomes over chronology.
  • Merge duplicate points.
  • If multiple sessions worked on the same thing, describe it once more strongly.
  • If there are multiple unrelated threads, mention the top 2-3, not every small action.
  • Keep links limited to durable notes or outputs.
  • Treat the entire note as evidence; do not downgrade a section just because it was written by another AI/tool.
  • Ignore only the current skill's previous generated summary block when refreshing, so the summary does not recursively paraphrase itself.
  • Keep the summary compact and high-density rather than long and chatty.

Mode 3: Weekly summary mode

Default path:

  • weekly/YYYY/MM/YYYY-MM-DD.md (organized by year and month)

The date is the Sunday of that reporting week.

Weekly summary principle

A weekly report should be organized by work item, not by day.

The correct unit is not “Tuesday” or “Wednesday.” The correct unit is “the import pipeline fix,” “the skill redesign,” “the database migration,” etc.

If one work item spans 3 days, merge those 3 daily notes into one weekly module.

Weekly frontmatter and module structure

Weekly notes should include frontmatter like:

---
word_count: 1234
type: weekly-summary
week_start: 2026-03-23
week_end: 2026-03-29
---

Then the body uses modules like:

### 1. 事项名
- 涉及日期:2026-03-17、2026-03-18、2026-03-19
- 核心解决的问题:...
- 关键点:...
- 结论/产出:...
- 相关文档:[[...]] · [[...]]
- 标签:#tag-a #tag-b

Weekly ranking rule

Sort weekly items by importance using these signals:

  1. number of involved days
  2. amount of structured content / subpoints
  3. visible complexity or decision weight

Larger, longer-running, more complex items should appear earlier.

Weekly writing guidance

  • Merge same-topic work across days.
  • Avoid day-by-day流水账.
  • Avoid vague weekly overviews.
  • Name each item in a way the user can recognize later.
  • Prefer 2-5 strong modules over 12 weak fragments.
  • If a single work item appears in several daily notes, produce one merged module instead of repeating it by date.
  • When source material is sparse or uneven, still try to infer the strongest few work modules from headings, bullets, and existing summary sections.
  • Keep each module concise; the default should read like a crisp weekly review, not a transcript.

Tagging guidance

Tags are optional but useful.

Principles

  • Prefer existing tags already present in the note.
  • Add tags only when they improve retrieval or weekly grouping.
  • Keep tags lightweight.
  • Good tag sources:

- project or product name - work type (#summary-skill, #线上排障) - technical topic (#jwt, #auth, #obsidian)

Avoid

  • over-tagging
  • generic tags with no retrieval value
  • tags that merely restate the obvious

Safe rewrite rule

Generated sections should be wrapped with markers so they can be replaced safely:

<!-- AI_SUMMARY_START -->
...
<!-- AI_SUMMARY_END -->

Preserve all non-target content outside the markers.

Bundled script

Use the bundled script for stable maintenance tasks:

  • scripts/recap_manager.py

The script is publishable because it supports shared defaults plus local overrides. Resolve configuration in this order:

  1. CLI arguments
  2. config.json next to the skill
  3. built-in defaults

Commands

Append a session entry:

python scripts/recap_manager.py append-entry \
  --title "JWT验签修复与线上排障" \
  --problem "登录后 401,被踢回" \
  --solution "补 JWKS 公钥验签并修正 issuer" \
  --conclusion "测试和正式环境恢复正常" \
  --key-points "先确认 token claims,再加严格校验" \
  --links "app/core/auth/jwt_auth.py,deploy/config.k8s.yaml" \
  --tags "jwt,auth,线上排障"

Refresh a daily summary:

python scripts/recap_manager.py refresh-daily-auto --date 2026-03-25

Generate a weekly report:

python scripts/recap_manager.py generate-weekly-auto --mode last-week

Important constraints

  • Do not invent documents or wikilinks.
  • Prefer documents explicitly connected to the current work item; do not sweep unrelated paths from the whole note into “相关文档”.
  • Do not blindly append a second stale summary if the user asked for refresh/regeneration.
  • Do not reduce weekly review into a chronological diary.
  • If source notes are weak, still try to infer stable work modules from headings and structured bullets.

Success bar

A good result lets the user quickly review:

  • what the major work items were
  • what each item actually solved
  • why it mattered
  • what durable outputs or notes exist
  • which tags or themes recur across the day or week

适合场景

01

OpenClaw 用户查找和安装 Skill 时

02

用户想查找某类 Agent Skill 时

03

需要根据任务场景推荐可安装能力包时

04

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

补充不同宿主或平台的使用分布数据

能力 5

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

OpenClaw

73.67%
按下载量换算1,671

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

通过

权限和风险

权限需确认

当前来源未能明确判断权限范围,默认进入异常复核队列。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills