Token导航 LogoToken导航

Agent Skill · 效率

peripheral-vision(周边视觉)

peripheral-vision 用于处理图像、截图、视觉识别或图片素材相关工作,适合在 OpenClaw 中需要让 Agent 分析图片、整理视觉素材或辅助图像流程时使用。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。

peripheral-vision 是效率类 Agent Skill,可用宿主为 OpenClaw,通过ClawHub获取,权限为接触敏感数据;许可证为 MIT-0,第三方扫描有提醒项,按安装量在效率类的 21,003 个 Skill 里排第 6,620。

总安装 9,096周安装 379GitHub Stars 公开资料未说明下载 3,032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
1.0.0

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:peripheral-vision(周边视觉)
来源仓库:https://github.com/jcools1977/peripheral-vision
安装命令:
openclaw skills install peripheral-vision
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 peripheral-vision

这个 Skill 是做什么的

效率类 Skill 把重复的整理、汇总、转换工作交给助手完成,减少手工复制粘贴。

peripheral-vision 用于处理图像、截图和视觉识别相关工作,适合分析图片或辅助图像流程。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。它监视相邻系统、上游依赖和下游使用者,以发现可能影响当前工作的变更。

涉及系统监控时应设置合理告警阈值和通知机制。

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 peripheral-vision;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

peripheral-vision 怎么安装?

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

peripheral-vision 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。

peripheral-vision 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。

peripheral-vision 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。

peripheral-vision 用的人多吗?

公开数据是累计安装 9,096 次,近一周安装 379 次,下载 3,032 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。

peripheral-vision 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

  • 需要接触敏感数据,装到常用环境前要先评估影响面。
  • 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。
  • 需要处理效率类任务、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。
  • 需要厂商级技术支持和 SLA 的生产系统。

效率类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
github

GitHub 代码协作 · OpenClaw

7,632,918545需要联网MIT-0通过
nano-pdf

Nano PDF 文档 · OpenClaw

4,249,476231需要联网MIT-0通过
auto-updater

自动更新程序 · OpenClaw

2,964,063386操作浏览器MIT-0通过
video-frames

视频帧 · OpenClaw

2,621,304112需要联网MIT-0通过
gemini

Gemini 助手 · OpenClaw

2,547,39648需要联网MIT-0有提醒
peripheral-vision

周边视觉 · 本页

9,096—接触敏感数据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
peripheral-vision
version
1.0.0
description
>
author
J. DeVere Cooley
category
situational-awareness
tags
metadata
openclaw
emoji
👁️
os
["darwin", "linux", "win32"]
cost
free
requires_api
false
tags

Peripheral Vision

"You don't crash because you weren't looking ahead. You crash because you weren't looking to the side."

What It Does

When you're deep in a module, your attention is foveal — focused on the code directly in front of you. But the code that breaks your work is rarely the code you're staring at. It's the code *next to* it: the upstream service that changed its response format, the downstream consumer that started sending unexpected input, the shared utility that someone refactored while you were mid-feature.

Peripheral Vision creates a situational awareness field around your current work context. It continuously scans the edges of your attention and alerts you to changes that are:

  1. Not in the files you're editing
  2. But in files that your files depend on, interact with, or assume stability from

The Awareness Model

Visual Field Anatomy

In human vision:

  • Foveal: The 2° center. Maximum detail. Where you're looking right now.
  • Parafoveal: The next 5°. Some detail. Aware but not focused.
  • Peripheral: Everything else. No detail. Detects motion and threats.

In code:

ZoneCode EquivalentWhat Peripheral Vision Does
FovealFiles you're actively editingNothing (you're already looking here)
ParafovealFiles directly imported/called by your filesMonitors for recent changes
PeripheralFiles that interact transitively — shared deps, consumers, upstream providersScans for changes that could cascade to you
Blind SpotFiles you have no awareness of but are connected through implicit channels (shared DB, env vars, runtime config)Attempts detection through co-change analysis

The Six Peripheral Channels

Channel 1: Upstream Drift

Monitors: Libraries, modules, and services your code imports or calls.

SCANS FOR:
├── Interface changes (new required params, removed fields, renamed exports)
├── Behavioral changes (same interface, different semantics)
├── Version bumps in dependencies that affect your used surface area
├── Deprecation warnings for functions you actively use
└── Changelog entries tagged "breaking" in your dependency tree

Channel 2: Downstream Pressure

Monitors: Code that imports, calls, or depends on your code.

SCANS FOR:
├── New consumers of your module you didn't know about
├── Consumers using your code in ways you didn't intend
├── Consumers that would break if you change your current interface
├── Test files in other modules that test your behavior (coupling signal)
└── Downstream code that reimplements your functionality (trust signal)

Channel 3: Sibling Mutations

Monitors: Other files in your module or directory that were recently changed by someone else.

SCANS FOR:
├── Changes to shared utilities, helpers, or constants you use
├── Changes to configuration files that affect your module's behavior
├── New files that might conflict with or duplicate your current work
├── Refactors that changed naming conventions your code follows
└── Changes to test infrastructure or fixtures your tests depend on

Channel 4: Schema Tremors

Monitors: Database schemas, API contracts, protobuf definitions, GraphQL types — any shared data shape.

SCANS FOR:
├── Column additions/removals in tables your code queries
├── Type changes in fields your code reads or writes
├── New constraints (NOT NULL, UNIQUE) that could cause your writes to fail
├── API version changes in services you consume
└── Protobuf/GraphQL field deprecations or renames

Channel 5: Environmental Shifts

Monitors: Infrastructure, configuration, and deployment context changes.

SCANS FOR:
├── Changes to environment variables your code reads
├── Changes to Docker/container configurations
├── CI/CD pipeline modifications that affect your build/test/deploy
├── Infrastructure changes (new regions, changed timeouts, updated limits)
└── Changes to shared tooling (linter rules, formatter settings, build configs)

Channel 6: Temporal Neighbors

Monitors: Files that historically change at the same time as your files (co-change analysis).

SCANS FOR:
├── Files that have changed in the same commit as your files > 3 times
├── Files that have changed within 24h of your files > 5 times
├── Files that have triggered the same CI failures as your files
└── Files that share the same bug-fix pattern as your files

How It Works

Phase 1: CONTEXT CAPTURE
├── Identify the foveal zone (files currently open/modified)
├── Trace parafoveal zone (direct dependencies, both import and export)
├── Map peripheral zone (transitive deps, shared resources, co-changers)
└── Identify blind spots (connected through implicit channels)

Phase 2: CHANGE DETECTION
├── For each zone, detect changes since your work started:
│   ├── Git: commits by others to files in your awareness field
│   ├── Schema: migrations or type definition changes
│   ├── Config: environment or infrastructure changes
│   └── Dependency: upstream version bumps or changelog entries
├── Classify each change by relevance to your current work
└── Score impact probability (how likely this affects you)

Phase 3: ALERT TRIAGE
├── Filter out noise (changes with < 20% impact probability)
├── Classify remaining by urgency:
│   ├── STOP: Your current work may be invalidated
│   ├── REVIEW: Your current work should account for this
│   ├── NOTE: Awareness-only, no action needed yet
│   └── EMERGING: Pattern forming, not yet actionable
└── Generate contextual alert with specific implications for your work

Phase 4: CONTINUOUS MONITOR
├── Re-scan periodically (configurable interval)
├── Update awareness field as your focus shifts
├── Accumulate pattern data for temporal neighbor analysis
└── Fade old alerts that are no longer relevant

Alert Format

╔══════════════════════════════════════════════════════════════╗
║               PERIPHERAL VISION ALERT                       ║
║          Context: You're working on src/checkout/            ║
╠══════════════════════════════════════════════════════════════╣
║                                                              ║
║  🔴 STOP (1)                                                 ║
║  ├── [Schema] payments table: 'currency' column type changed ║
║  │   from varchar(3) to enum — your INSERT will fail         ║
║  │   Changed by: @devB in migration 2025_03_01_alter_pmts    ║
║  │   Action: Update PaymentBuilder to use enum values        ║
║                                                              ║
║  🟡 REVIEW (2)                                               ║
║  ├── [Upstream] CartService.getTotal() now returns            ║
║  │   {amount, currency} instead of just amount               ║
║  │   Changed by: @devC in commit a8f3d2e (2h ago)            ║
║  │   Action: Destructure correctly or your amounts are objects║
║  │                                                           ║
║  ├── [Sibling] src/checkout/utils.ts was refactored          ║
║  │   formatPrice() renamed to formatCurrency()               ║
║  │   Changed by: @devA in commit b9c4e1f (4h ago)            ║
║  │   Action: Update your imports before pushing              ║
║                                                              ║
║  🔵 NOTE (1)                                                 ║
║  ├── [Temporal] test/fixtures/cart-data.json was updated     ║
║  │   This file co-changes with checkout/ 73% of the time     ║
║  │   Your checkout tests may need updated fixtures            ║
║                                                              ║
╚══════════════════════════════════════════════════════════════╝

When to Invoke

  • When starting a new feature branch (establish your awareness field)
  • Before committing (check what changed around you while you were focused)
  • Before opening a PR (ensure your work accounts for parallel changes)
  • After pulling from main (understand what landed near your code)
  • When a CI build fails unexpectedly (was it your change or a peripheral one?)

Why It Matters

The most frustrating bugs aren't the ones you introduced. They're the ones that were introduced *next to* you, while you were focused on your own work, in code you had no reason to look at. By the time you discover them, you've built on top of broken assumptions.

Peripheral Vision doesn't replace careful code review. It gives your code review peripheral awareness — so you're not just reviewing what changed, but what changed *around* what changed.

Zero external dependencies. Zero API calls. Pure git and static analysis.

平台分布

OpenClaw

87.91%
按下载量换算2,665

安全审计

VirusTotal

通过

ClawScan

可疑

Static analysis

通过

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。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