Skip to content

Latest commit

 

History

2 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Echo Journal · 思想回声日志

为仍愿意独立思考的人,留下一处可以慢慢回看的地方。
A quiet place to write slowly, read closely, and keep thinking.


About · 关于

让记录不止于保存

Echo Journal 是一个可安装给 AI Agent 使用的日记与反思 Skill。它不替你总结人生,也不急于把感受归类为问题或答案;它更像一位有立场的陪读者,在你的文字旁留下回声:什么判断已经成立,什么推论还需要证据,什么知识线索值得继续追问。

在 AI 可以迅速生成文字、意见与结论的时代,人仍需要保有一种缓慢的能力:分辨实际观察、据此作出的解释,以及希望发生的事。Echo Journal 想见证这种能力如何在时间里生长——不是替人思考,而是督促人不放弃思考。

Let a journal become more than an archive

Echo Journal is an installable journaling and reflection skill for AI agents. It does not summarise a life or rush feelings into problems and answers. It reads alongside you with a position: noticing what already holds, where an inference may have moved ahead of its evidence, and what knowledge is worth pursuing next.

In an age when AI can produce language, opinions, and conclusions at speed, we still need a slower human capacity: to distinguish what we observed, what we inferred, and what we hoped might happen. Echo Journal witnesses that capacity over time—not to think in anyone’s place, but to make it harder to abandon independent thought.

“世界上最好的工作是终身学习者在自由市场中的创造性表达。”

“The best work in the world is the creative expression of a lifelong learner in a free market.”

Write slowly · Read closely · Keep thinking


Features · 功能

1. Echo|单篇日记回应

对用户的日记生成一段 120–220 字的连续回应,而不是总结或鸡汤。它会:

  • 区分事实/感受、解释与愿望;
  • 保留文本中已有依据的判断;
  • 只挑战解释或愿望,不否定感受;
  • 明确指出推理跳跃、替代解释与还需要的观察;
  • 仅在能改变判断边界时加入哲学、社会学、科学、伦理学或经济学的“知识侧光”;
  • 附一条可选的“反向检验”,以及必有的 3–5 个拓展学习标签。

Generate a 120–220-character continuous response to one journal entry—not a summary or generic encouragement. Echo will:

  • distinguish facts / lived feelings, interpretations, and wishes;
  • preserve a judgment that is already supported by the text;
  • challenge only an interpretation or wish, never invalidate a feeling;
  • name a reasoning leap, an alternative explanation, and the observation still needed;
  • bring in a philosophical, sociological, scientific, ethical, or economic side-light only when it changes the boundary of the judgment;
  • add an optional reverse-test question and 3–5 mandatory learning hashtags.

2. Reflect|阶段性反思

比较一组日记,识别反复出现的问题、判断转折与证据标准的变化。所有主题都必须指向具体日志 ID;AI 的长期判断需要用户确认后才可保存。

Compare a set of entries to identify recurring questions, turning points, and changing standards of evidence. Every theme must cite specific entry IDs; longer-term AI interpretations are saved only after user confirmation.

3. Recall|长期回溯

回答“我过去怎么理解亲密关系?”这类问题。它从已授权的全部历史 Markdown 日志中找出最相关的 3–6 篇,不受时间远近限制,并比较:

  • 不变的立场;
  • 变化的判断;
  • 证据标准的变化;
  • 仍未解决的问题。

Answer questions such as “How did I understand intimacy in the past?” by retrieving the 3–6 most relevant authorized Markdown entries, regardless of recency. It compares:

  • durable positions;
  • changed judgments;
  • changing standards of evidence;
  • unresolved questions.

4. Export|完整 Markdown 档案

将已保存的全部历史日记和全部 Echo 按日期与时间顺序合并为一个 Markdown 文件。每篇保留原文、回应、反向检验、拓展学习、检索说明和来源。

Merge every saved historical entry and its Echo into one Markdown archive, ordered by date and time. Each entry preserves the original text, response, reverse test, learning prompts, research note, and sources.

Installation · 安装

Codex

将整个 echo-journal/ 文件夹复制到 Codex 的 skills 目录(通常为 ~/.codex/skills/),重新打开或刷新 Codex 后即可使用。

Copy the entire echo-journal/ folder into Codex’s skills directory (usually ~/.codex/skills/), then reopen or refresh Codex.

其他 Agent

将本仓库中的 SKILL.mdreferences/ 提供给支持 Markdown Skill / Prompt 包的 Agent。该 Skill 需要 Agent 具备:

  • 读取与写入用户明确授权的本地 Markdown 文件的能力;
  • 在用户要求知识背景时进行联网检索并给出来源的能力;
  • 尊重用户隐私与文件保存位置的能力。

For other agents, provide this repository’s SKILL.md and references/ directory to an agent that supports Markdown skills or prompt packages. The agent should be able to:

  • read and write local Markdown files only in user-authorized locations;
  • browse and cite sources when the user asks for knowledge context;
  • respect privacy and the user’s chosen storage path.

Usage · 使用方式

直接用自然语言发起即可:

Start in natural language:

这是我今天的日记:……
Here is my journal entry for today: …
请回溯:我过去怎么理解亲密关系?
Recall: How did I understand intimacy in the past?
请对我 2026 年 6 月的日志做一次 reflect。
Please reflect on my journal entries from June 2026.
请将我此前全部已保存的日志和回声导出为一个 Markdown 文件,保存到……
Export all my previously saved journals and Echoes into one Markdown file at …

首次保存时,Agent 会要求用户明确指定日记目录。推荐结构:

Before the first save, the agent asks the user to name a journal directory. Recommended layout:

echo-journal/
├── journal/       # original entries and Echoes / 每篇原始日志与 Echo
├── memory/        # user-confirmed themes and processing log / 经确认的主题与记录
└── exports/       # merged Markdown archives / 合并后的完整 Markdown 档案

Principles & Boundaries · 原则与边界

  • 用户原文优先:不擅自改写日记。

  • 不模仿真实人物、不虚构引语、不把思想家当作权威化身。

  • 不进行心理诊断;不把文学表达还原为对作者的心理结论。

  • 每个不同意都必须说明:原判断为何合理、何处超出证据、何种新观察可能改变结论。

  • 外部研究只说明一般规律,不能裁定一段具体关系、一次目光或一个选择的真实含义。

  • 所有持久化文件必须由用户授权位置后才可写入。

  • User text comes first: never silently rewrite a journal.

  • Do not imitate real people, fabricate quotations, or use thinkers as authority figures.

  • Do not diagnose; do not reduce literary language to a psychological conclusion about its author.

  • Every disagreement must explain why the original judgment is plausible, where it exceeds available evidence, and what new observation could change it.

  • External research can describe general patterns, but cannot decide the meaning of one particular relationship, gaze, or choice.

  • Write persistent files only after the user authorizes a location.

Privacy · 隐私

Echo Journal 的长期回溯依赖用户自己保存的 Markdown 日志。它只应读取用户明确授权的目录;不要将私人日志、未确认的长期判断或本地绝对路径提交到公开仓库。

Long-term recall relies on Markdown entries saved by the user. Echo Journal should read only explicitly authorized directories; never commit private journals, unconfirmed long-term interpretations, or local absolute paths to a public repository.

v1.0.0 Release Notes · 发布说明

首个公开版本包含 Echo、Reflect、Recall 与完整 Markdown Export;支持思想透镜、联网知识侧光、反向检验、拓展学习标签和跨日记的证据标准追踪。

The first public release includes Echo, Reflect, Recall, and full Markdown export, plus perspective cards, source-grounded knowledge side-lights, reverse tests, learning hashtags, and cross-entry tracking of evidence standards.

License · 许可证

本作品采用 Creative Commons Attribution-NonCommercial 4.0 International(CC BY-NC 4.0)许可协议。

允许非商业性使用、转载、改编与分享;请注明作者、作品来源、许可协议链接,并标明修改情况。禁止将本作品用于商业目的。详见 LICENSE

This work is licensed under Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0). Non-commercial use, sharing, and adaptation are allowed with appropriate credit, a link to the license, and an indication of changes. Commercial use is prohibited. See LICENSE.

About

A source-grounded journaling skill for independent thought, reflection, and long-term recall.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors