Token导航 LogoToken导航TokenDH.com
研究检索需要联网github未标认证来源可访问许可证需确认审计提醒

desloppifydesloppify 搜索

Agent Skill

desloppify 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

432

周安装

18

GitHub Stars

公开资料未说明

下载量

144
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:desloppify(desloppify 搜索)
来源仓库:https://github.com/fasalzein/desloppify
仓库路径:skills/desloppify
安装命令:
npx skills add https://github.com/fasalzein/desloppify --skill desloppify
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/fasalzein/desloppify --skill desloppify

简介

用于查找、检索和筛选相关信息,适合快速定位候选结果。

  • 适用于需要根据关键词、任务场景或来源线索进行信息定位的场景。
  • 支持代码质量扫描、死代码清理和结构化重构审核流程。
  • 安装前建议确认权限范围、维护状态及是否会触发联网或文件读写操作。
  • desloppify 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Desloppify

Yes — keep this file lean.

This SKILL.md is the operational entrypoint. Put only the workflow and decision points here. Keep detailed command/config/reference material in companion docs and load them only when needed:

  • reference/commands.md
  • reference/packs-and-tools.md
  • reference/config-and-plugins.md
  • reference/overhaul-review.md

What this skill is for

Use it when the user wants to:

  • desloppify a repo
  • clean up AI slop
  • run a code-quality pass
  • scan for dead code / weak types / complexity / anti-patterns
  • run a structured overhaul review or phased cleanup audit

The CLI detects. The agent triages and decides what to fix.

Default behavior is still scan → triage → fix. Only switch into overhaul-review mode when the user clearly wants a broader audit or phased cleanup plan.

Default workflow

Optional overhaul-review mode

If the user wants a broader audit instead of an immediate cleanup pass, load reference/overhaul-review.md and choose a mode first:

  • Surgical — one theme only
  • Systematic — section-by-section review, pause at each section
  • Full audit — broad phased roadmap

Do not silently switch into overhaul mode for an ordinary desloppify request.

0) Set up

Run:

desloppify check-tools [path]

If the repo clearly needs setup help, also use:

desloppify install-skill
desloppify install-hooks
desloppify setup

install-hooks scaffolds .githooks/ into the current git repo and enables it. The installed hooks resolve the CLI from the repo, bunx, or PATH; auto-pick the repo's suggested pack when unambiguous; and refuse to clobber unmanaged hook files or replace another active hook chain from .git/hooks, .husky, or worktree-scoped hook config. Hook runs default to current changes; set DESLOPPIFY_HOOK_SCOPE=repo when the user wants hook scans to cover the whole repo.

If you need the exact commands or pack/tool setup details, read:

  • reference/commands.md
  • reference/packs-and-tools.md

1) Scan

Start with a normal scan for the most relevant pack.

Common default:

desloppify scan [path] --pack js-ts

That default JS/TS path intentionally keeps madge off the main run so missing or slow circular-deps analysis does not block ordinary scans. Only add --with-madge or --category circular-deps when the user explicitly wants circular dependency analysis.

Then read artifacts in this order:

  1. .desloppify/reports/latest.findings.json
  2. desloppify report [path]
  3. .desloppify/reports/latest.report.md
  4. .desloppify/reports/latest.wiki.json or .desloppify/reports/latest.handoff.md

The first artifact-writing scan also auto-adds .desloppify/ to .gitignore in git repos so this state stays local by default.

Use --json, --markdown, --wiki, --handoff, --delta, or benchmark commands only when the task needs those outputs. On very large repos, prefer --json --summary so stdout stays compact while the full artifact is still written to disk. For exact variants, read reference/commands.md.

2) Triage

For each issue bucket, decide:

  • Fix — worth changing now
  • Skip — intentional or low-value churn
  • Flag-only — public API / dynamic behavior / serialization / risky refactors

Important:

  • do not mistake external analyzer warnings for a clean repo
  • partial scans skip whole-project analyzers
  • full JS/TS scans skip madge by default unless you opt in
  • check-tools can suggest among the shipped js-ts, python, rust, go, and ruby packs before you scan

If you need pack/tool behavior details, read reference/packs-and-tools.md.

3) Fix

Prefer isolated worktrees for parallel fix streams:

desloppify worktrees [path]

Then give each fix agent:

  1. the worktree path
  2. the saved report path or a focused scan slice
  3. desloppify fix. --safe first
  4. a re-scan after edits

Current rule of thumb:

  • fix --safe is the dependable default
  • broader fix tiers are partial and need judgment

4) Verify

After fixes:

desloppify scan [path]
desloppify report [path]

Use desloppify delta... when you need regression gating or before/after comparison.

When to load the companion docs

Load reference/commands.md when you need:

  • exact CLI invocations
  • report / delta / benchmark flows
  • artifact paths

Load reference/packs-and-tools.md when you need:

  • supported packs
  • tool adapters
  • install commands
  • madge / partial-scan / repo-local tool behavior

Load reference/config-and-plugins.md when you need:

  • config file discovery
  • supported config fields
  • plugin examples
  • desloppify/plugin-api
  • plugin rules.<id>.options
  • the stable root programmatic API
  • identityGroup / fix plugin behavior

Load reference/overhaul-review.md when you need:

  • Surgical / Systematic / Full review modes
  • sectioned audit flow
  • impact/effort matrix output
  • out-of-scope / failure-mode / execution-order summaries

Guidance for large findings

Treat GOD_FILE as a refactor signal, not a blind split command. Inspect why it tripped:

  • too many imports
  • mixed route + DB concerns
  • barrel/export concentration
  • too many HTTP methods in one route file
  • too many useState calls
  • generic bucket-file structure

Safety tiers

  • T1 — mechanical fixes
  • T2 — AST-validated fixes
  • T3 — cross-file fixes
  • Flag-only — never auto-fix

Suppression

Inline:

console.log("intentional"); // desloppify:ignore CONSOLE_LOG

Project-level suppression uses .desloppifyignore.

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

35.22%
按下载量换算51

Claude

31.04%
按下载量换算45

Cursor

20.18%
按下载量换算29

Gemini CLI

10.74%
按下载量换算15

安全审计

Gen Agent Trust Hub

可疑

Socket

通过

Snyk

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

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