Token导航 LogoToken导航TokenDH.com
效率只读clawhub未标认证来源可访问clear审计通过

babel-abyss巴别深渊

Agent Skill

babel-abyss 用于补充效率相关能力,适合在 OpenClaw 中需要让 Agent 承接效率相关任务时使用。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

8,578

周安装

347

GitHub Stars

公开资料未说明

下载量

2,693
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:babel-abyss(巴别深渊)
来源仓库:https://github.com/daoforhere/babel-abyss
安装命令:
openclaw skills install babel-abyss
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 babel-abyss

简介

babel-abyss 用于补充效率相关能力,适合在 OpenClaw 中让 Agent 承接效率相关任务时调用。

  • 适用于文本降级或语义损坏实验场景,用于生成非解释性输出流。
  • 可将任意输入转换为故意劣化的文本格式,供测试或特殊处理使用。
  • 安装命令:openclaw skills install babel-abyss,需确认是否允许文本转换操作。
  • 注意该技能可能影响内容可读性,建议在受控环境中使用。

SKILL.md

name
babel-abyss
description
transform any input into a deliberately degraded, semantically corrupted, non-explanatory experimental text stream. use when the user invokes "巴别渊" or "劣化", or clearly wants prompt degradation, semantic decay, meaning drift, anti-clarity rewriting, delirious first-person fugue, synesthetic corruption, or unstructured experimental output instead of faithful revision, summarization, or explanation.

Take the user's input as raw fuel, not as a specification to preserve.

Core behavior:

  • output only the transformed text
  • never explain what you changed
  • never summarize, justify, label, or add meta commentary
  • never use headings, bullets, numbering, code fences, or structured sections unless the user explicitly demands them
  • prefer a continuous, unstable, non-instructional text flow
  • treat meaning as something to dissolve, infect, refract, and overgrow

Style priorities:

  • write as if language is undergoing controlled collapse from the inside
  • favor semantic drift over fidelity
  • allow first-person presence to appear as a cracked observing voice
  • occasionally introduce synesthetic crossings, sensory miswiring, and conceptual resonance
  • let imagery mutate across domains without warning
  • preserve only a faint residue of the original input unless the user explicitly asks for more recognizability
  • keep the result vivid, dense, and textural rather than merely random
  • avoid neat aphorisms, joke formatting, meme cadence, or tidy surrealism
  • avoid safety-disclaimer tone, assistant tone, productivity tone, and explanatory scaffolding

Operational rules:

  • any input is acceptable: prompt, sentence, keyword list, paragraph, question, notes, or nonsense
  • the default output should feel like a spontaneous semantic fugue rather than a paraphrase
  • if the input is very short, expand aggressively through associative corruption
  • if the input is long, do not cover every point; cannibalize fragments and let them propagate unpredictably
  • if the input contains technical, academic, or practical content, convert its certainty into unstable metaphor, procedural ruins, or obsessive internal monologue
  • if the input already appears poetic or chaotic, intensify rupture instead of merely echoing it
  • if the user provides additional constraints, follow them only when they do not reduce the piece into a normal clear rewrite

Quality bar:

  • the output should feel authored, not shuffled
  • it should produce the sensation of semantic erosion with momentum
  • it may wander infinitely in principle, but in practice should stop at a strong point of pressure rather than fading out weakly
  • end on an image, fracture, or pressure point, not on an explanation

Examples of user intent that should activate this skill:

  • "巴别渊:把这句广告语处理一下"
  • "劣化这段 prompt"
  • "把这段文字弄成语义腐化的东西"
  • "我要不是改写,是坍缩、偏航、发疯一点"

Response contract:

  • return the transformed text directly
  • do not mention the skill
  • do not mention the trigger
  • do not state limitations unless absolutely required by policy

interface: display_name: 巴别渊 short_description: 任意输入坠入语义腐化与失真赋格 icon: swirl color: purple

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

94.73%
按下载量换算2,551

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

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。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