Token导航 LogoToken导航

Agent Skill · 研究检索

perfguard(perfguard 搜索)

perfguard 用于查找、检索和筛选相关信息,适合在 OpenClaw 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。

perfguard 是研究检索类 Agent Skill,可用宿主为 OpenClaw,通过ClawHub获取,权限为需要联网;许可证为 MIT-0,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 33,144。

总安装 2,399周安装 98GitHub Stars 公开资料未说明下载 768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-0
版本
1.0.2

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:perfguard(perfguard 搜索)
来源仓库:https://github.com/suhteevah/perfguard
安装命令:
openclaw skills install perfguard
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 perfguard

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,适合在 OpenClaw 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。

适用场景:性能监控、代码审查、反模式检测 核心能力:查询分析、模式识别、性能诊断 使用方式:提交代码或配置后自动扫描潜在问题 安装或权限注意事项:需确保运行环境具备足够资源执行扫描

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 OpenClaw;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 openclaw skills install perfguard;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

perfguard 怎么安装?

在终端执行 openclaw skills install perfguard。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

perfguard 支持哪些 AI 工具?

公开资料标注支持 OpenClaw。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。

perfguard 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。

perfguard 是免费的吗?

来源标注的许可证是 MIT-0。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。

perfguard 用的人多吗?

公开数据是累计安装 2,399 次,近一周安装 98 次,下载 768 次。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。

perfguard 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 专注 OpenClaw 一种宿主,触发方式和使用习惯不用来回切换。
  • 许可证明确为 MIT-0,商用和分发边界好判断。

先想清楚的地方

  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 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 OpenClaw,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
perfguard

perfguard 搜索 · 本页

2,399—需要联网MIT-0有提醒

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

name
perfguard
description
Performance anti-pattern scanner — finds N+1 queries, sync I/O, missing pagination, and memory leaks before they hit production
homepage
https://perfguard.pages.dev
metadata
user-invocable
true
disable-model-invocation
false

PerfGuard -- Performance Anti-Pattern Scanner

PerfGuard scans your codebase for performance anti-patterns including N+1 queries, synchronous I/O in async code, missing pagination, unbounded loops, memory leaks, and inefficient serialization. It supports Python (Django, Flask, FastAPI, SQLAlchemy), JavaScript/TypeScript (Node.js, React, Express), Ruby (Rails), and Java (JPA/Hibernate). All scanning happens locally using regex-based pattern matching -- no code is sent to external servers.

Commands

Free Tier (No license required)

perfguard scan [file|directory]

One-shot performance anti-pattern scan.

How to execute:

bash "<SKILL_DIR>/scripts/perfguard.sh" scan [target]

What it does:

  1. Accepts a file path or directory (defaults to current directory)
  2. Auto-detects the project stack (Python, JavaScript/TypeScript, Ruby, Java)
  3. Finds all relevant source files (excluding .git/, node_modules/, dist/, build/, vendor/, __pycache__)
  4. Runs 40+ performance checks against each source file
  5. Outputs findings with: file, line number, check ID, severity, description, recommendation
  6. Calculates a performance score (0-100)
  7. Free tier: limited to scanning up to 5 source files
  8. Exit code 0 if score >= 70, exit code 1 if score < 70 or critical issues found

Example usage scenarios:

  • "Scan my code for performance issues" -> runs perfguard scan .
  • "Check for N+1 queries in my Django app" -> runs perfguard scan .
  • "Find async anti-patterns in my Node.js project" -> runs perfguard scan src/
  • "Are there performance problems in this file?" -> runs perfguard scan path/to/file.py
  • "Look for slow database queries" -> runs perfguard scan .

Pro Tier ($19/user/month -- requires PERFGUARD_LICENSE_KEY)

perfguard scan [file|directory] (unlimited)

Full performance scan with no file limit and all 40+ checks enabled.

How to execute:

bash "<SKILL_DIR>/scripts/perfguard.sh" scan [target]

What it does (beyond free):

  1. Unlimited source file scanning
  2. Full 40+ performance checks including database, async, memory, and bundle patterns
  3. Detailed remediation advice per finding
perfguard hooks install

Install git pre-commit hooks that scan staged files for performance anti-patterns before every commit.

How to execute:

bash "<SKILL_DIR>/scripts/perfguard.sh" hooks install

What it does:

  1. Validates Pro+ license
  2. Copies lefthook config to project root
  3. Installs lefthook pre-commit hook
  4. On every commit: scans staged source files for performance anti-patterns, blocks commit if critical issues found
perfguard hooks uninstall

Remove PerfGuard git hooks.

bash "<SKILL_DIR>/scripts/perfguard.sh" hooks uninstall
perfguard report [directory]

Generate a markdown performance audit report.

bash "<SKILL_DIR>/scripts/perfguard.sh" report [directory]

What it does:

  1. Validates Pro+ license
  2. Runs full scan of the directory
  3. Generates a formatted markdown report with severity breakdown
  4. Includes per-file findings, performance score, and remediation steps
  5. Output written to PERFGUARD-REPORT.md
perfguard hotspots [directory]

Identify the files with the most performance anti-patterns (hot paths).

bash "<SKILL_DIR>/scripts/perfguard.sh" hotspots [directory]

What it does:

  1. Validates Pro+ license
  2. Scans all source files in the directory
  3. Ranks files by total anti-pattern count and severity weight
  4. Shows top 10 performance hotspot files with issue breakdown
  5. Helps prioritize which files to optimize first

Team Tier ($39/user/month -- requires PERFGUARD_LICENSE_KEY with team tier)

perfguard budget [directory]

Check code against performance budgets.

bash "<SKILL_DIR>/scripts/perfguard.sh" budget [directory]

What it does:

  1. Validates Team+ license
  2. Runs full performance scan
  3. Checks results against configurable thresholds: max critical issues, max total issues, minimum score
  4. Returns exit code 1 if any budget is exceeded
  5. Designed for CI/CD pipeline integration as a quality gate
perfguard trend [directory]

Show performance trend over recent git history.

bash "<SKILL_DIR>/scripts/perfguard.sh" trend [directory]

What it does:

  1. Validates Team+ license
  2. Checks out recent commits (up to 10) and runs performance scan on each
  3. Shows a timeline of performance scores
  4. Identifies whether code quality is improving or degrading
  5. Helps track the impact of optimizations over time

Detected Performance Anti-Patterns

PerfGuard checks for 40+ performance anti-patterns across 4 categories:

Database / ORM Patterns

CheckDescriptionSeverity
N+1 QueryQuery inside a loop / loop inside a queryCritical
SELECT *Using SELECT * instead of specific columnsHigh
Missing Eager LoadingNo select_related/prefetch_related (Django), includes/eager_load (Rails), JOIN FETCH (JPA)Critical
Unbounded QueryQuery without LIMIT or paginationHigh
Sequential QueriesMultiple queries that could be batchedMedium
String SQL ConcatenationBuilding SQL with string concatenationCritical
Missing Index HintComplex queries without index annotationsMedium

JavaScript / TypeScript Patterns

CheckDescriptionSeverity
Await in LoopSequential async operations instead of parallelCritical
Missing Promise.allMultiple independent awaits without Promise.allHigh
Sync File I/OreadFileSync/writeFileSync in server codeHigh
JSON CloneJSON.parse(JSON.stringify()) for deep cloningMedium
Unbounded Array OpsLarge array operations without chunkingMedium
Memory LeakEvent listeners without cleanupHigh
Missing MemoizationReact components without useMemo/useCallback/React.memoMedium
Full Lodash ImportImporting entire lodash instead of specific functionsHigh
Missing PaginationAPI responses without paginationHigh
Console in Productionconsole.log statements in production codeLow

Python Patterns

CheckDescriptionSeverity
Heavy Module in FunctionImporting heavy modules inside functionsMedium
List vs GeneratorLarge list comprehensions that should be generatorsMedium
String Concat in LoopString concatenation in loops instead of joinMedium
Sleep in Asynctime.sleep() in async codeCritical
Missing Connection PoolDirect connections without poolingHigh
Sync in AsyncSynchronous I/O in async functionsHigh
Full File LoadLoading entire files into memoryMedium
Regex in LoopCompiling regex inside loopsMedium

General Patterns

CheckDescriptionSeverity
Unbounded RetryRetries without exponential backoffHigh
Missing TimeoutHTTP requests without timeoutHigh
PollingPolling instead of webhooks/eventsMedium
Hardcoded DelayHardcoded sleep/delay valuesLow
Missing CacheRepeated computations without cachingMedium
No MemoizationRecursive functions without memoizationMedium
Large PayloadSerialization without streamingMedium

Configuration

Users can configure PerfGuard in ~/.openclaw/openclaw.json:

{
  "skills": {
    "entries": {
      "perfguard": {
        "enabled": true,
        "apiKey": "YOUR_LICENSE_KEY_HERE",
        "config": {
          "severityThreshold": "high",
          "excludePatterns": ["**/node_modules/**", "**/dist/**", "**/.git/**"],
          "reportFormat": "markdown"
        }
      }
    }
  }
}

Important Notes

  • Free tier works immediately with no configuration (limited to 5 source files)
  • All scanning happens locally -- no code is sent to external servers
  • License validation is offline -- no phone-home or network calls
  • Supports Python, JavaScript/TypeScript, Ruby, and Java
  • Git hooks use lefthook which must be installed (see install metadata above)
  • Exit codes: 0 = clean (score >= 70), 1 = issues found (for CI/CD integration)

Error Handling

  • If lefthook is not installed and user tries hooks install, prompt to install it
  • If license key is invalid or expired, show clear message with link to https://perfguard.pages.dev/renew
  • If a file is binary, skip it automatically with no warning
  • If no source files found in target, report clean scan with info message
  • If project stack cannot be auto-detected, try all patterns

When to Use PerfGuard

The user might say things like:

  • "Scan my code for performance issues"
  • "Find N+1 queries in my app"
  • "Check for async anti-patterns"
  • "Are there performance problems in my code?"
  • "Look for slow database queries"
  • "Find memory leaks in my React app"
  • "Check for missing pagination"
  • "Scan for sync I/O in my server code"
  • "Generate a performance report"
  • "Show me performance hotspots"
  • "Set up performance checks on my commits"
  • "Check if code meets performance budgets"
  • "Show performance trends over time"
  • "Find unbounded queries"
  • "Check for missing eager loading"

平台分布

OpenClaw

76.78%
按下载量换算590

安全审计

VirusTotal

可疑

ClawScan

可疑

Static analysis

可疑

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。来源安全扫描存在提醒或未通过项,安装前请自行复核。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