Token导航 LogoToken导航

Agent Skill · 研究检索

add-badges(添加徽章)

add-badges 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。

add-badges 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为需要联网;许可证信息还不明确,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 42,966。

总安装 1,024周安装 44GitHub Stars 2下载 359核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

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:add-badges(添加徽章)
来源仓库:https://github.com/wyattowalsh/agents
仓库路径:skills/add-badges
安装命令:
npx skills add https://github.com/wyattowalsh/agents --skill add-badges
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/wyattowalsh/agents --skill add-badges

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

add-badges 用于自动生成和更新项目 README 中的徽章(badges)信息。适合在 Codex、Claude、Cursor、Gemini CLI 中需要快速展示项目状态、包版本或 CI 状态时使用。通过 GitHub 安装,需确认项目路径和徽章类型,支持 dry-run 预览避免误改文件。

安装前建议检查现有徽章块,避免重复插入或格式冲突。注意该技能依赖 Claude hooks,可能自动执行代码,需谨慎授权后再使用。

来源仓库内路径:skills/add-badges

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/wyattowalsh/agents --skill add-badges;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要需要联网,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

add-badges 怎么安装?

在终端执行 npx skills add https://github.com/wyattowalsh/agents --skill add-badges。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

add-badges 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

add-badges 需要什么权限?

权限级别是需要联网。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。

add-badges 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

add-badges 用的人多吗?

公开数据是累计安装 1,024 次,近一周安装 44 次,下载 359 次,GitHub 2 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

add-badges 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
add-badges

添加徽章 · 本页

1,0242需要联网未明确通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Add Badges

Generate and update README badge blocks without drifting into general README editing.

Dispatch

$ARGUMENTSAction
EmptyShow the empty/help gallery and ask for the target repo or README path
Flags or natural-language badge requestRun the full badge workflow: detect -> select -> present -> insert
--dry-run with any requestRun through preview only and stop before file edits
removal-only requestSkip detection, inspect the existing badge block, and present the reduced set

Empty/Help Gallery

When $ARGUMENTS is empty, show the main usage shape plus examples:

  • /add-badges --profile active --dry-run
  • /add-badges --include status,package --readme README.md
  • /add-badges --replace --yes
  • /add-badges remove social badges

State the main phases and the write boundary:

  • Detect the repo and existing badge state
  • Select badges and layout based on flags and detected signals
  • Present the proposed block and diff indicators
  • Insert only after approval, unless --yes was explicitly passed

Progressive Disclosure

  • Keep SKILL.md focused on routing, phase order, approval boundaries, and flag semantics.
  • Read references/badge-catalog-core.md on every run.
  • Read references/badge-catalog-extended.md only when non-basic signals are detected.
  • Read references/style-guide.md only when layout, format-specific syntax, or dark-mode output matters.
  • Use scripts/detect.py for repository detection and scripts/validate-badges.py only after insertion when verification is requested.

Internal Phase 1 — Detect

Run the detection script via the Bash tool:

uv run python skills/add-badges/scripts/detect.py <path>

Parse the JSON output. The script detects: repo info, languages, package managers, frameworks, CI/CD, infrastructure, code quality, testing, docs, license, release, security, community, developer tooling, databases, monorepo signals, and existing badges.

If the script fails (uv unavailable, Python missing, script error, invalid JSON), fall back to manual detection:

  • Glob for manifest files (pyproject.toml, package.json, go.mod, Cargo.toml, etc.)
  • Read .git/config or run git remote get-url origin for owner/repo
  • Glob .github/workflows/*.yml for CI badge candidates
  • Read LICENSE first line for SPDX type
  • Check for Dockerfile,.pre-commit-config.yaml, codecov.yml, etc.

Internal Phase 2 — Select

Read references/badge-catalog-core.md (always). Read references/badge-catalog-extended.md when detection reports any non-basic signals (any of: frameworks, infrastructure, code_quality linters/formatters/type_checkers, docs, release, security, monorepo, databases, developer_tooling, community).

Read references/style-guide.md for layout, ordering, and URL conventions.

Selection rules:

  • Prefer dynamic endpoints over static — never hardcode version numbers, coverage %, download counts
  • Include ?logo={slug}&logoColor={white|black} on every badge with a Simple Icons slug
  • Match existing badge style if README already has badges (from existing_badges.style); default to flat-square
  • If platform is GitLab or Bitbucket, use shields.io platform paths (/gitlab/..., /bitbucket/...)
  • --profile <name>: preset badge selection by maturity. new (3-5 badges: status, license, language), active (8-12: core + quality, code-style, frameworks), mature (12-18: core + all extended except developer-tooling), enterprise (15-20: ALL 16 categories including OpenSSF). Overrides --include/--exclude. No profile = auto-select based on detected features
  • Target 8-15 badges; group by display super-groups: Status > Quality > Package > Tech Stack > Social

Display super-group → category mapping:

Super-groupCategories
Statusstatus
Qualityquality, code-style, security
Packagepackage, license
Tech Stacklanguage, frameworks, infrastructure, docs, release, databases, monorepo, developer-tooling
Socialsocial, community
  • Deduplicate against existing badges by comparing badge URL service/metric paths
  • For CI status: prefer native badge URL (e.g., github.com/{owner}/{repo}/actions/workflows/{file}/badge.svg) — works for private repos. For repos with 5+ workflows, prioritize CI/test workflows; let user select others in Phase 3
  • If repo.visibility is "private", skip badges marked requires: public-api and warn user. Prefer native URLs for CI and direct service URLs for coverage
  • If existing badges reference dead services (from existing_badges.dead_services), flag them and suggest catalog replacements
  • Custom badge support: if user requests a badge not in the catalog (Discord server, sponsor, custom API), construct from shields.io static badge API or endpoint badge API (/badge/dynamic/json?url=...&query=...). Ask for required params
  • If user requests forthebadge.com-style decorative badges, use forthebadge.com API. Note: forthebadge is decorative only — no dynamic data. For dynamic badges in large bold style, use shields.io ?style=for-the-badge

Flag handling:

  • --include <categories>: only generate badges from named categories (comma-separated). Category names: status, quality, package, license, language, social, code-style, frameworks, infrastructure, docs, release, databases, monorepo, community, security, developer-tooling. Mutually exclusive with --exclude — error if both provided
  • --exclude <categories>: skip named categories. Same names as --include. Also accepts display group alias: tech-stack expands to language,frameworks,infrastructure,docs,release,databases,monorepo,developer-tooling. Mutually exclusive with --include
  • --style <style>: override badge style (flat, flat-square, plastic, for-the-badge, social)
  • --layout <layout>: badge arrangement — inline (default), centered, grouped, table, collapsible. collapsible wraps secondary groups (Tech Stack, Social) in <details><summary> elements — ideal for 15+ badges. See style-guide.md
  • --readme <path>: target a specific file instead of auto-detected README
  • --dark-mode: generate <picture> elements with <source media="(prefers-color-scheme: dark)"> for theme-aware badges on GitHub. Produce HTML instead of pure markdown
  • --dry-run: output proposed badge block and diff without modifying any file. Exit after preview
  • --yes: skip approval prompt before modifying files
  • --replace: replace content within markers AND consolidate scattered badges into the marker block

Internal Phase 3 — Present

Show grouped preview with category headers. Render badges as actual [![alt](url)](link) markdown so the user can see them.

Diff indicators when updating existing badges:

  • [+] new badge being added
  • [=] existing badge being kept
  • [-] existing badge being removed

If scattered badges exist outside markers, show their locations and offer to consolidate into the marker block.

If user requests removal only (e.g., "remove social badges"), skip detection, read existing badge block, remove specified badges, present updated block.

Offer to reorder, add, or remove individual badges before finalizing. Accept natural language adjustments ("move stars before license", "drop the forks badge", "add a Discord badge").

Ask for approval before modifying files. Skip approval if --yes passed. Never skip prompts for missing required info (owner/repo, workflow file names, etc.) even with --yes.

If --dry-run, output the proposed block and exit without modifying files.

Internal Phase 4 — Insert

Find existing <!-- BADGES:START --> / <!-- BADGES:END --> markers or create them.

Insert approved badges grouped by display super-group (Status, Quality, Package, Tech Stack, Social). Separate groups with a blank line. Add <!-- generated by add-badges YYYY-MM-DD --> comment inside the marker block.

Insertion point priority:

  1. Existing markers (replace content between them)
  2. After first # Title heading
  3. Top of file if no heading found

If no README exists, create one with # {repo-name} heading then add badges.

If --dark-mode, wrap each badge in <picture> elements per style-guide.md.

--replace consolidates any scattered badges found outside markers into the marker block. Show user exactly what will be moved — nothing silently deleted.

Preserve any manual content outside markers.

After insertion, optionally run uv run python skills/add-badges/scripts/validate-badges.py <readme-path> via Bash to verify all badge URLs return valid responses. Report any broken or slow badges to the user.

Stop Hooks

Stop and ask instead of writing when any of these are true:

  • the target README path is unclear
  • required repo identity details are missing for dynamic badges
  • workflow file names or badge-specific parameters are missing
  • the user did not approve a write and --yes was not passed
  • --replace would move or consolidate scattered badges the user has not seen yet

For write operations:

  • --dry-run must always stop after preview and diff
  • approval is mandatory before insertion unless --yes is explicit
  • missing required info never gets bypassed by --yes
  • preserve all non-marker README content and stop if the requested change would require broader README editing

Style Rules

Default style: flat-square. Match existing style if badges already present. Use for-the-badge for hero/landing sections. Use social style for star/follow count badges. Separate display groups with a blank line in the output.

Error Handling

  • No git remote: prompt for owner/repo. If user declines, generate only static language/framework/tooling badges (no remote-dependent badges)
  • No manifest files: static-only badges
  • RST/AsciiDoc README: use appropriate image syntax (see style-guide.md for format-specific syntax)
  • Mixed existing styles: recommend standardizing, ask user which style
  • Style-only change (--style with no other changes): preserve current badge set, update style param only
  • Private repo detected: skip requires: public-api badges, prefer native badge URLs, warn user
  • detect.py produces partial results: work with what is available, note missing sections

Tricky Icon Slugs

Common Simple Icons gotchas: gnubash not bash, nodedotjs not node, vuedotjs not vue, nextdotjs not next, .env is dotenv, springboot not spring-boot, flydotio not fly. Note: nuxt is now correct (was nuxtdotjs). For icons not in the catalog, use lowercase brand name; check simpleicons.org if unsure.

Golden Example

<!-- BADGES:START -->
<!-- generated by add-badges 2025-01-15 -->
[![CI](https://github.com/{owner}/{repo}/actions/workflows/ci.yml/badge.svg)](https://github.com/{owner}/{repo}/actions)
[![PyPI](https://img.shields.io/pypi/v/{package}?style=flat-square&logo=pypi&logoColor=white)](https://pypi.org/project/{package}/)
[![License](https://img.shields.io/github/license/{owner}/{repo}?style=flat-square)](https://github.com/{owner}/{repo}/blob/main/LICENSE)
[![OpenSSF Scorecard](https://api.scorecard.dev/projects/github.com/{owner}/{repo}/badge)](https://scorecard.dev/viewer/?uri=github.com/{owner}/{repo})
<!-- BADGES:END -->

Reference File Index

FileRead When
references/badge-catalog-core.mdAlways — provides base badge definitions for common categories
references/badge-catalog-extended.mdDetection reports non-basic signals (frameworks, infrastructure, linters, docs, etc.)
references/style-guide.mdDetermining layout, ordering, URL conventions, and format-specific syntax

Canonical Vocabulary

Canonical TermMeaning
badgeA shields.io (or similar) image element displaying project metadata
shieldSynonym for badge; the rendered SVG from shields.io
styleVisual variant: flat, flat-square, plastic, for-the-badge, social
super-groupDisplay grouping: Status, Quality, Package, Tech Stack, Social
categoryClassification of a badge type (e.g., license, language, frameworks)
markerHTML comments (<!-- BADGES:START/END -->) delimiting the badge block
slugSimple Icons identifier used in ?logo= parameter (e.g., gnubash)
profilePreset badge selection by project maturity: new, active, mature, enterprise
dynamic badgeBadge that fetches live data from an endpoint (version, coverage, downloads)
static badgeBadge with hardcoded text — avoid when a dynamic alternative exists

Scope Boundaries

IS for: badge detection, badge selection, badge preview, badge insertion inside marker blocks, and badge-specific README updates.

NOT for: general README rewriting, broader docs work, CI/CD design, or non-badge documentation changes.

Critical Rules

Non-negotiable constraints for every badge operation:

  1. Never hardcode dynamic values. Always use shields.io dynamic endpoints for versions, coverage, and download counts.
  2. Always include logo parameters. Every badge with a Simple Icons slug must have ?logo={slug}&logoColor={white|black}.
  3. Respect existing style. Match the style of badges already present in the README; default to flat-square only when no badges exist.
  4. Never modify content outside markers. Preserve all manual content outside <!-- BADGES:START --> / <!-- BADGES:END --> markers.
  5. Always ask before writing. Never modify the README without user approval unless --yes was explicitly passed.
  6. Check icon slugs carefully. Use the Tricky Icon Slugs list; verify against simpleicons.org when uncertain.
  7. Skip private-incompatible badges. When repo.visibility is "private", never generate badges marked requires: public-api.

平台分布

Codex

38.02%
按下载量换算136

Claude

30.05%
按下载量换算108

Cursor

19.76%
按下载量换算71

Gemini CLI

10.03%
按下载量换算36

安全审计

Gen Agent Trust Hub

通过

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。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills