← 返回 Skills 市场
deaiify
作者
shawnpetros
· GitHub ↗
· v3.4.1
· MIT-0
112
总下载
0
收藏
1
当前安装
7
版本数
在 OpenClaw 中安装
/install deaiify
功能描述
Ban LLM em-dashes and en-dashes before delivery, rewrite via embedded LLM.
使用说明 (SKILL.md)
deAIify Plugin
Enforces the "no em dashes" rule across all output by intercepting messages, detecting U+2013 (en dash) and U+2014 (em dash), and forcing a rewrite turn via embedded LLM correction.
What it does
- Detects U+2013 and U+2014 in outbound messages
- Calls embedded LLM to rewrite without dashes (preserves meaning, tone, style)
- Returns corrected output to user
- Never touches U+002D (hyphen-minus in code)
- Excludes content inside code blocks and inline code
How it works
before_agent_replyintercepts the completed assistant text before delivery- Detects banned dash characters outside of code blocks
- Calls
runEmbeddedPiAgentwith a restructuring prompt - Verifies the rewrite is sane (word count and length checks)
- Returns the rewritten reply
Installation
openclaw plugin install deaiify
Configuration
| Option | Type | Default | Description |
|---|---|---|---|
rewriteTimeoutMs |
integer | 15000 |
Timeout in ms for the embedded rewrite call |
The plugin uses the session's default model for rewrites. No model config needed.
安全使用建议
This plugin appears to do exactly what it claims: intercept replies containing em-dashes/en-dashes and ask the session's embedded model to rewrite them before delivery. Before installing, consider: (1) any assistant output passed to the embedded model (the prompt + original text) will be sent to your configured model provider — avoid installing if that could leak PII or secrets; (2) the embedded run is given workspace/agent directory context and a scoped sessionKey which is normal for embedded runs but increases what the rewrite agent can see; (3) the plugin fails open (delivers original text) on errors and has recursion protections. If those trade-offs are acceptable, the plugin is coherent and proportional. If you are worried about sensitive content being forwarded, do not enable it or restrict it to sessions where no sensitive data is present.
能力标签
能力评估
Purpose & Capability
Name/description match the implementation. The code and SKILL.md both implement detection of U+2013/U+2014 outside code blocks, call an embedded rewrite run, verify the output, and return the rewritten reply. No unrelated env vars, binaries, or external services are requested.
Instruction Scope
SKILL.md accurately describes the runtime flow and the code follows it. Note: the plugin passes workspaceDir, agentDir and a sessionKey to runEmbeddedPiAgent and writes a short-lived sessionFile (deleted in finally). That means the embedded agent run may have access to agent workspace/context and the session key scope — behaviour that is reasonable for an embedded rewrite but worth being aware of.
Install Mechanism
No external download/install spec is declared; runtime is delivered as plugin code. Package files are present but there is no installer that fetches arbitrary third-party binaries or URLs. No extract-from-unknown-URL or similar high-risk install steps are present.
Credentials
The plugin requests no environment variables or credentials. It does, however, send the assistant text (the full prompt+text) to the session's configured model via runEmbeddedPiAgent and forwards model/provider identifiers and (when present) a sessionKey scoped with a deaiify suffix. If the message text contains sensitive data you do not want sent to your configured model provider, this is the primary privacy consideration.
Persistence & Privilege
always is false, the plugin does not assert permanent presence or modify other plugins. It creates a temporary session transcript file and attempts to delete it on completion. No system-wide settings or other skills' configs are modified.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install deaiify - 安装完成后，直接呼叫该 Skill 的名称或使用
/deaiify触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v3.4.1
Fix VT scanner flag: replace mkdtemp+rm-rf dropper pattern with flat tmp file + unlink
v3.4.0
v3 rewrite: proper em-dash detection with code-block exclusion, embedded LLM rewrite
v3.0.3
Version 3.0.3 — Major cleanup
- Removed 17 files, including all source code, build outputs, tests, configuration, and documentation files.
- No functional changes or feature updates.
v3.0.2
v3.0.2: Fixed version bump, LLM rewrite engine for dash removal
v3.1.0
Fix YAML frontmatter rendering on ClawHub page, restore Billy Mays infomercial README
v3.0.1
Fix SKILL.md frontmatter rendering on ClawHub page
v3.0.0
Summary: Version 3.0.0 introduces automated enforcement to remove em-dashes and en-dashes from all outbound messages.
- Intercepts outgoing messages to detect U+2013 (en dash) and U+2014 (em dash)
- Automatically rewrites such messages via embedded LLM, preserving meaning and style
- Ensures no visible formatting changes for users
- Standard hyphens (U+002D) in code remain unaffected
- Activation requires configuration in `openclaw.json`
元数据
常见问题
deaiify 是什么？
Ban LLM em-dashes and en-dashes before delivery, rewrite via embedded LLM. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 112 次。
如何安装 deaiify？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install deaiify」即可一键安装，无需额外配置。
deaiify 是免费的吗？
是的，deaiify 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
deaiify 支持哪些平台？
deaiify 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 deaiify？
由 shawnpetros（@shawnpetros）开发并维护，当前版本 v3.4.1。
推荐 Skills