vision
Analyze Images with a Vision Model
Text-only agents cannot inspect screenshots, diagrams, or scanned pages. This skill sends one image and a focused prompt to your configured vision model.
安装前请停下并请求确认。
请审查安装计划，并在更改文件前获得用户明确同意。
使用我的智能体安装
将此请求复制给你的智能体，其中包含权威 Skill 页面和清单。
Review the Skillstore skill "vision" from https://skillstore.io/skills/kotot-vision.md and its manifest at https://skillstore.io/api/skills/kotot-vision/manifest. Verify the artifact. Stop and obtain explicit user consent before installing or changing files.你的智能体仍应展示其计划，并按安全策略请求必要确认。
Agent 可读资源
当 AI Agent、爬虫或脚本需要干净上下文，而不是读取完整页面时，请使用这些链接。
测试它
正在使用“vision”。 Describe a dashboard screenshot with a failed deployment.
预期结果:
The deployment status is Failed. The visible error cites a missing database connection variable in the production environment.
正在使用“vision”。 Transcribe a photographed receipt.
预期结果:
- Merchant: North Market
- Date: 2026-08-14
- Total: $42.80
- The item description on the second line is unclear.
正在使用“vision”。 Explain an architecture diagram.
预期结果:
The client sends requests through an API gateway to two services. Both services share a database, while the worker also reads from a queue.
安全审计
高风险The two outbound HTTP findings are confirmed because the script posts image data and prompts to a configurable endpoint. Most other findings are documentation, placeholders, or restricted configuration access, but inherited project settings can silently redirect uploads.
已确认安全问题 (1)
能力复核项 (2)
这些是真实的本地能力，对此技能可能是预期行为，因此需要复核，但不会按已确认恶意行为计分。
风险因素
⚙️ 外部命令 (27)
🌐 网络访问 (11)
📁 文件系统访问 (15)
分享与引用此报告
分享版本化评估报告、中性徽章、嵌入卡片和引用。Skillstore 只报告证据，不判断该 Skill 是否安全。
复制报告链接
https://skillstore.io/skills/kotot-vision/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdown 徽章
[](https://skillstore.io/skills/kotot-vision?utm_source=security_passport_badge)HTML 徽章
<a href="https://skillstore.io/skills/kotot-vision?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/kotot-vision/security.svg" alt="Skillstore security assessment" loading="lazy"></a>嵌入卡片
<iframe src="https://skillstore.io/embed/skills/kotot-vision.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>学术引用 (APA · BibTeX · CFF)
APA 引用
kotot. (2026). vision security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/kotot-vision/audits/1BibTeX 引用
@techreport{kotot-kotot-vision-2026,
author = {kotot},
title = {vision security audit report (audit version 1)},
institution = {Skillstore},
year = {2026},
number = {1},
url = {https://skillstore.io/skills/kotot-vision/audits/1},
note = {Author version unspecified}
}CITATION.cff
cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "vision security audit report (audit version 1)"
version: "unspecified"
type: report
authors:
- name: "kotot"
date-released: "2026-08-16"
url: "https://skillstore.io/skills/kotot-vision/audits/1"
identifiers:
- type: other
value: "skillstore:kotot-vision:audit:1"
description: "Skillstore immutable audit report identifier"
比较变体
3 个可安装变体每位作者的技能都保持为独立的可安装项。推荐变体按 Skillstore 证据排序。
为什么此变体排名第一
Skillstore 评分
评分依据 证据置信度: 中你能构建什么
Diagnose Screenshot Errors
Read visible error messages and identify likely causes from application screenshots.
Extract Scanned Text
Transcribe text from receipts, forms, and scanned pages while preserving useful structure.
Interpret Visual Research
Summarize charts and diagrams, including labels, values, relationships, and visible caveats.
试试这些提示
Analyze [image path or URL]. Describe the main subjects, setting, visible text, and important details.
Read [image path or URL]. Transcribe all visible text and preserve headings, lists, and table relationships.
Inspect [screenshot path]. Identify the exact error, affected component, likely cause, and evidence visible in the interface.
Analyze [image path or URL] against [expected state]. List matches, discrepancies, uncertain details, and evidence for each conclusion.
最佳实践
- Use a specific question that names the details you need.
- Verify the resolved endpoint before sending confidential images.
- Ask the model to mark unreadable or uncertain content.
避免
- Do not send secrets or regulated data to an untrusted endpoint.
- Do not treat OCR output as authoritative without verification.
- Do not request broad descriptions when exact values are required.
常见问题
Which image sources are supported?
Does this skill require a specific provider?
Is an API key required?
Where is configuration loaded from?
Can I inspect a request before sending it?
Does the skill guarantee accurate OCR?
开发者详情
作者
kotot许可证
MIT
Skillstore 修订
r1
版本提示
作者未声明版本。
引用
d7ec88d380bc0952a475899451d682383ab5d76c
维护新鲜度
2026/8/16
使用量
0 次下载 · 0 次浏览