Token导航 LogoToken导航

Agent Skill · 研究检索

detector(detector 搜索)

detector 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。

detector 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,权限为接触敏感数据;许可证为 MIT,暂无第三方扫描结果,按安装量在研究检索类的 73,675 个 Skill 里排第 59,000。

总安装 396周安装 17GitHub Stars 127下载 139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
MIT

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:detector(detector 搜索)
来源仓库:https://github.com/anton-abyzov/specweave
仓库路径:skills/detector
安装命令:
npx skills add https://github.com/anton-abyzov/specweave --skill detector
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。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/anton-abyzov/specweave --skill detector

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。

用于查找、检索和筛选相关信息,适合快速定位候选结果。可结合关键词、任务场景或来源线索进行信息聚合。通过命令行安装并使用,需参考原始 README 获取具体指令。安装前建议确认是否会触发联网或文件读写,确保权限可控。detector 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

来源仓库内路径:skills/detector

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/anton-abyzov/specweave --skill detector;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

    它需要接触敏感数据,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

detector 怎么安装?

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

detector 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。

detector 需要什么权限?

权限级别是接触敏感数据。该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

detector 是免费的吗?

来源标注的许可证是 MIT。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

detector 用的人多吗?

公开数据是累计安装 396 次,近一周安装 17 次,下载 139 次,GitHub 127 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。

detector 安全吗?

没有查到第三方扫描结果,安全性无法从公开资料判断,安装前请自行审查源码、依赖和命令行为。

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。
  • 3 个目录同时收录,说明这个 Skill 流传比较广。
  • 许可证明确为 MIT,商用和分发边界好判断。

先想清楚的地方

  • 需要接触敏感数据,装到常用环境前要先评估影响面。
  • 没有第三方扫描结果,安全性只能靠自己读源码判断。
  • 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

  • 环境里有大量密钥、凭证,不方便给第三方 Skill 读取的人。
  • 要求有安全扫描结论才肯安装的团队。
  • 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
detector

detector 搜索 · 本页

396127接触敏感数据MIT未收录

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

SpecWeave - Smart Workflow Documentation

SpecWeave provides explicit slash commands for reliable workflow execution.

NOTE: Auto-detection of product descriptions is handled by the increment-planner skill. This skill provides command documentation and workflow guidance.

How SpecWeave Works

To use SpecWeave: Type /inc "Feature description" to start

Smart workflow features:

  • ✅ Auto-resume (/do finds next incomplete task)
  • ✅ Auto-close (/inc closes previous if PM gates pass)
  • ✅ Progress tracking (/progress shows status anytime)
  • ✅ Natural flow (finish → start next, no overhead)

Available Slash Commands

Core Workflow Commands

CommandAliasDescriptionExample
/increment/incPlan Increment (PM-led, auto-closes previous)/inc "User auth"
/do-Execute tasks (smart resume, hooks after every task)/do
/progress-Show status (task %, PM gates, next action)/progress
/validate-Validate quality (rule-based + optional LLM judge)/validate 0001 --quality
/done-Close explicitly (optional, /inc auto-closes)/done 0001

Supporting Commands

CommandDescriptionExample
/list-incrementsList all increments with status/list-increments
/sw:sync-docsReview strategic docs vs code/sw:sync-docs --increment=003
/sync-githubSync increment to GitHub issues/sync-github

Why Only ONE Alias?

Design decision: /inc is the ONLY alias (most frequently used command).

  • ✅ Minimizes cognitive overhead (one alias to remember)
  • ✅ Other commands use full names for clarity
  • ✅ Simpler mental model

Typical Workflow

Natural append-only workflow (0001 → 0002 → 0003):

# 1. Initialize project (CLI, before Claude session)
npx specweave init my-saas

# 2. Plan your first increment (PM-led)
/inc "AI-powered customer support chatbot"
# PM creates: spec.md + plan.md + tasks.md (auto!) + tests.md

# 3. Build it (smart resume)
/do
# Auto-resumes from next incomplete task
# Hooks run after EVERY task

# 4. Check progress anytime
/progress
# Shows: 5/12 tasks (42%), next: T006, PM gates status

# 5. Continue building
/do
# Picks up where you left off

# 6. Start next feature (auto-closes previous!)
/inc "real-time chat dashboard"
# Smart check:
#   PM gates pass → Auto-close 0001, create 0002
#   PM gates fail → Present options (never forces)

# 7. Keep building
/do
# Auto-finds active increment 0002

# Repeat: /sw:increment → /sw:do → /sw:progress → /sw:increment (auto-closes) → /sw:do...

Command Details

/inc or /increment - Plan Increment

Most important command! PM-led planning with auto-close intelligence.

# Short form (recommended)
/inc "User authentication with JWT and RBAC"

# Full form
/increment "User authentication with JWT and RBAC"

What happens:

  1. Smart Check Previous: If increment in-progress:

- PM gates pass → Auto-close previous, create new (seamless) - PM gates fail → Present options (complete first / move tasks / cancel)

  1. PM-Led Planning: PM Agent analyzes requirements
  2. Creates: spec.md (WHAT & WHY), plan.md (HOW)
  3. Auto-generates: tasks.md (from plan), tests.md (test strategy)
  4. Ready to build: Status set to "planned"

/do - Execute Tasks (Smart Resume)

Smart resume: Automatically finds next incomplete task.

# Auto-finds active increment, resumes from next task
/do

# Or specify increment explicitly
/do 0001

What happens:

  1. Finds active increment (or uses specified ID)
  2. Parses tasks.md, finds first incomplete task
  3. Shows resume context (task T006, description, priority)
  4. Executes task implementation
  5. Runs hooks after EVERY task completion (docs update, validation)
  6. Repeats for next task when you run /do again

No manual tracking needed! Just keep running /do.

/progress - Show Status

Progress visibility: See exactly where you are anytime.

/progress

# Auto-finds active increment, shows:
# - Task completion % (P1 weighted higher)
# - PM gates preview (tasks, tests, docs)
# - Next action guidance
# - Time tracking & stuck task warnings

/validate - Validate Quality

Two-level validation: Rule-based (120 checks) + optional AI quality judge.

# Rule-based validation only
/validate 0001

# With AI quality assessment (LLM-as-judge)
/validate 0001 --quality

# Export suggestions to tasks.md
/validate 0001 --quality --export

# Auto-fix issues (experimental)
/validate 0001 --quality --fix

/done - Close Explicitly

Optional command: Use when you need explicit closure (usually /inc handles this).

/done 0001

# System validates:
# - All P1 tasks completed
# - All tests passing
# - Documentation updated
#
# Offers leftover transfer options for P2/P3 tasks

When to use:

  • Explicit closure before long break
  • Force closure without starting new increment
  • Generate closure report only

Usually NOT needed: /inc auto-closes previous increment if PM gates pass.

/list-increments - List All

WIP tracking: View all increments with status and completion.

# All increments
/list-increments

# Filter by status
/list-increments --status in-progress

# Filter by priority
/list-increments --priority P1

# Show task breakdown
/list-increments --verbose

# Only WIP increments
/list-increments --wip-only

Smart Workflow Features

1. Auto-Resume (No Manual Tracking)

Problem: Traditional workflows require manual tracking ("which task am I on?")

Solution: /do automatically finds next incomplete task.

/do

📋 Resuming increment 0001-authentication
   Next: T006 - Implement JWT token validation
   Priority: P1
   Estimate: 2 hours
   Context: After T005 (token generation)

Starting task T006...

2. Auto-Close (Seamless Flow)

Problem: Manual closure overhead ("do I need to close this?")

Solution: /inc auto-closes previous if PM gates pass.

Happy path (auto-close):

/inc "payment processing"

📊 Checking previous increment 0001-authentication...
   PM Gates: ✅ All P1 complete, tests pass, docs updated

✅ Auto-closing 0001 (seamless)
Creating 0002-payment-processing...

Issues found (present options):

/inc "payment processing"

📊 Checking previous increment 0001-authentication...
   PM Gates: ❌ 2 P1 tasks remaining

❌ Cannot auto-close 0001 (incomplete)

Options:
  A) Complete 0001 first (recommended)
  B) Move incomplete tasks to 0002
  C) Cancel new increment

Your choice? _

3. Suggest, Never Force

Critical principle: User always in control.

  • ✅ Present options when issues found
  • ✅ Explain consequences clearly
  • ✅ Let user decide
  • ❌ NEVER surprise user with forced closure

4. Progress Visibility

Problem: Status unclear ("how much is done?")

Solution: /progress shows status anytime.

/progress

📊 Increment 0001-authentication

Status: in-progress
Progress: 42% (5/12 tasks) ⏳

Task Breakdown:
  P1: 60% (3/5) ⏳
  P2: 33% (2/6)
  P3: 0% (0/1)

PM Gates Preview:
  ✅ All P1 tasks: 60% (not ready)
  ⏳ Tests passing: Running...
  ✅ Docs updated: Yes

Next Action: Complete T006 (P1, 2h)
Time on increment: 3 days

Why Slash Commands?

Problem: Auto-activation doesn't work reliably in Claude Code.

SpecWeave solution: EXPLICIT slash commands for 100% reliability.

Benefits:

  • ✅ 100% reliable activation (no guessing)
  • ✅ Clear user intent (explicit action)
  • ✅ Consistent behavior (no surprises)
  • ✅ Easy to learn (visible in.claude/commands/)

How to Get Help

Within Claude Code:

User: "How do I use SpecWeave?"
→ Claude shows this documentation

Available commands:

User: "What SpecWeave commands are available?"
→ Claude lists all slash commands

Command syntax:

User: "How do I create a new increment?"
→ Claude explains /sw:increment command with examples

Documentation


💡 Pro Tip: Master the smart workflow cycle!

Core cycle: /inc (plan) → /do (implement) → /progress (check) → /inc (next)

Key insight: Natural flow without overhead. Focus on building, not project management.

One alias to remember: /inc (short for /increment)

Project-Specific Learnings

Before starting work, check for project-specific learnings:

# Check if skill memory exists for this skill
cat .specweave/skill-memories/detector.md 2>/dev/null || echo "No project learnings yet"

Project learnings are automatically captured by the reflection system when corrections or patterns are identified during development. These learnings help you understand project-specific conventions and past decisions.

平台分布

Claude Code

26.31%
按下载量换算37

Antigravity

23.24%
按下载量换算32

Cursor

19.01%
按下载量换算26

Gemini CLI

12.18%
按下载量换算17

OpenCode

7.56%
按下载量换算11

Codex

3.37%
按下载量换算5

安全审计

暂无安全审计结果可展示。

权限和风险

接触敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

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。

继续浏览同类 Skills