Token导航 LogoToken导航TokenDH.com
研究检索只读github未标认证来源可访问许可证需确认审计异常

plain-text纯文本

Agent Skill

plain-text 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。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

256

周安装

11

GitHub Stars

37

下载量

90
CodexClaudeCursorGemini CLI

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

unknown

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:plain-text(纯文本)
来源仓库:https://github.com/simhacker/moollm
仓库路径:skills/plain-text
安装命令:
npx skills add https://github.com/simhacker/moollm --skill plain-text
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 npx skills 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。

skills.shnpx skills
npx skills add https://github.com/simhacker/moollm --skill plain-text

简介

plain-text 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。

  • 它支持通过关键词、任务场景或来源仓库进行信息检索与筛选,帮助 Agent 快速获取相关资源。
  • 可通过 npx skills add 命令从指定 GitHub 仓库安装并使用该技能。
  • 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操作。
  • plain-text 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Plain Text

*"I don't want to worry about whatever cursed format OneNote uses still being something I can extract in 2035."* — @Havoc, Hacker News (2025)

What Is It?

Plain text is MOOLLM's substrate. Everything important is stored in human-readable text files:

  • SESSION.md — Session logs in Markdown
  • ROOM.yml — Room state in YAML
  • CHARACTER.yml — Character data in YAML
  • SKILL.md — Skill documentation in Markdown

No proprietary formats. No binary blobs. No databases. Just text.


Why Plain Text Wins

1. Durability

FormatStatus in 2004Status in 2025Status in 2045?
Microsoft Word.docUbiquitousLegacyUnknown
OneNoteNewCurrentUnknown
Markdown.mdNewUbiquitousStill readable
Plain.txtAncientUniversalStill readable

Text files from 1970 are still readable. They will be readable in 2070.

2. No Lock-In

# This file requires:
# - Any text editor
# - That's it

No vendor. No license. No subscription. No "cloud service discontinued."

3. Git-Friendly

- player_location: start
+ player_location: maze

Plain text diffs are:

  • Readable — Humans understand the change
  • Mergeable — Git can combine concurrent edits
  • Auditable — History is meaningful

4. LLM-Native

From Anil Dash (2025):

*"All of it — all of it — is controlled through Markdown files. When you see the brilliant work shown off from somebody who's bragging about what they made ChatGPT generate for them... all of the most advanced work has been prompted in Markdown."*

LLMs:

  • Are trained on plain text (GitHub, web, books)
  • Output plain text naturally
  • Understand structure in text (headers, indentation, bullets)
  • Process text more efficiently than structured formats

5. Inspectable

cat SESSION.md     # Read it
grep "player" *.yml # Search it
diff old.yml new.yml # Compare it
git log SESSION.md  # Track it

No special tools. No viewers. No converters.


The Unix Philosophy

Ken Thompson and Dennis Ritchie knew this in 1969:

*"Text is the universal interface."*

Pipes, streams, filters — all work on text:

cat rooms/*.yml | grep "exits" | sort | uniq

MOOLLM inherits this: the filesystem IS the database.


Plain Text Formats in MOOLLM

FormatExtensionPurpose
Markdown.mdProse, documentation, session logs
YAML.ymlStructured data with comments
Templates.tmplEmpathic templates (YAML + placeholders)

All are plain text. All are human-readable. All work with standard tools.


The "Worse is Better" Principle

Richard Gabriel (1989):

*"Simplicity is the most important consideration in a design."*

Markdown is "worse" than DocBook:

  • Fewer features
  • Less precise
  • Less formally specified

But Markdown won because:

  • Easier to learn — 5 minutes
  • Easier to read — Raw or rendered
  • Easier to write — No closing tags
  • Easier to implement — Hundreds of parsers

Simple formats that actually get used beat complex formats that don't.


Plain Text and AI

The entire trillion-dollar AI industry runs on plain text:

AI ComponentFormat
Training dataText files (books, web, code)
PromptsMarkdown / plain text
OutputsText / Markdown
Context windowsMeasured in text tokens
RAG documentsOften Markdown
Agent instructionsMarkdown files

From Anil Dash:

*"The trillion-dollar AI industry's system for controlling their most advanced platforms is a plain text format one guy made up for his blog."*

MOOLLM's Commitment

1. Files as State

No databases. State is in files:

adventure-4/
├── ADVENTURE.yml      # Game state
├── start/
│   ├── ROOM.yml       # Room state
│   └── lamp.yml       # Object state
└── characters/
    └── don-hopkins/
        └── CHARACTER.yml  # Character state

2. Human-Readable Always

If a human can't read the file in vim, rethink the format.

3. Comments ARE Data

# CRITICAL: Do not delete this
# It contains the player's memories
memories:
  - "Met Bumblewick in the maze"

Comments are preserved, not stripped.

4. No Transformation Gap

The YAML you write is the YAML that runs. No compilation step hides meaning.


Anti-Patterns

Binary formats — Opaque, require special tools ❌ Proprietary formats — Lock-in, potential for obsolescence ❌ Database-only state — Not inspectable, not git-friendly ❌ Generated-only files — If no human source exists, fragile ❌ Minified output — Readable formats should stay readable


Exceptions

Some things aren't text:

  • Images — Binary, but referenced from text
  • Audio/Video — Binary, but metadata in YAML
  • Large datasets — May need binary for performance

For these: keep the metadata and references in plain text, even if the content is binary.


Dovetails With


Protocol Symbol

PLAIN-TEXT

Invoke when: Choosing formats, designing storage, considering durability.


The Bottom Line

*"Markdown files from 2004 are still readable today. They'll be readable in 2045. Plain text is forever."*

Write for the future. Write in plain text.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

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

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

Codex

39.25%
按下载量换算35

Claude

30.07%
按下载量换算27

Cursor

19.12%
按下载量换算17

Gemini CLI

9.12%
按下载量换算8

安全审计

Gen Agent Trust Hub

未通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

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。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。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