📦
hol-guard
72Protect AI Harnesses with HOL Guard
Local AI harnesses can expose risky tools, plugins, and configuration changes. HOL Guard guides controlled setup, approval review, evidence collection, and package verification.
安全與法遵 查看
技能 plugin-scanner
📦
plugin-scanner
內容修訂 r1 中風險 ⚙️ 外部命令🌐 網路存取
Scan AI Plugins Before Installation
Unreviewed agent extensions can contain prompt injection, unsafe commands, secret exposure, or supply-chain risks. This skill guides local scanning and clear interpretation before trust.
支援: Claude Codex Code(CC)
使用我的智慧代理安裝
將此請求複製給你的智慧代理，其中包含權威 Skill 頁面和清單。
智能體請求
Review the Skillstore skill "plugin-scanner" from https://skillstore.io/skills/hashgraph-online-plugin-scanner.md and its manifest at https://skillstore.io/api/skills/hashgraph-online-plugin-scanner/manifest. Verify the artifact. You may proceed after verification, subject to the environment's own policy.你的智慧代理仍應顯示其計畫，並依安全政策請求必要確認。
Agent 可讀資源
當 AI Agent、爬蟲或腳本需要乾淨脈絡、而不是讀取完整頁面時，請使用這些連結。
測試它
正在使用「plugin-scanner」。 Scan the skill at ./skills/calendar-helper before installation.
預期結果:
The scan found one high-severity prompt-injection rule in SKILL.md. Review the cited instruction and remove the override language before installation.
正在使用「plugin-scanner」。 Verify the MCP server in ./integrations/search-server.
預期結果:
- Structure validation passed.
- The scan found a medium-severity command-execution pattern in the startup script.
- Review the cited command arguments and dependency source before use.
正在使用「plugin-scanner」。 Check ./plugins/reporter and tell me whether it is safe.
預期結果:
No covered issue was detected by the current scan. This result is limited to supported rules and does not guarantee safety.
安全審計
中風險v1 • 2026/8/27 開啟版本化報告
All 21 static findings are false positives caused by Markdown formatting, safety prohibitions, or documentation links. The workflow does present one real supply-chain risk because it proposes installing an unpinned package from PyPI. No prompt injection or secret-access intent was found.
1
已掃描檔案
96
分析行數
0
複核項目
0
已忽略的誤報
已確認的安全問題 (1)
中
Unpinned Third-Party Package Installation
The workflow proposes installing plugin-scanner from PyPI after approval. A compromised or replaced package could execute installation code within the isolated environment.
Lines 40-44 directly instruct pipx installation without a version pin or provenance check, while requiring user approval and isolation.
風險因素
審計單位: codex
分享與引用此報告
分享版本化評估報告、中性徽章、嵌入卡片與引用。Skillstore 只報告證據，不判斷該 Skill 是否安全。
複製報告連結
https://skillstore.io/skills/hashgraph-online-plugin-scanner/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdown 徽章
[](https://skillstore.io/skills/hashgraph-online-plugin-scanner?utm_source=security_passport_badge)HTML 徽章
<a href="https://skillstore.io/skills/hashgraph-online-plugin-scanner?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/hashgraph-online-plugin-scanner/security.svg" alt="Skillstore security assessment" loading="lazy"></a>嵌入卡片
<iframe src="https://skillstore.io/embed/skills/hashgraph-online-plugin-scanner.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>學術引用 (APA · BibTeX · CFF)
APA 引用
hashgraph-online. (2026). plugin-scanner security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/hashgraph-online-plugin-scanner/audits/1BibTeX 引用
@techreport{hashgraph-online-hashgraph-online-plugin-scanner-2026,
author = {hashgraph-online},
title = {plugin-scanner security audit report (audit version 1)},
institution = {Skillstore},
year = {2026},
number = {1},
url = {https://skillstore.io/skills/hashgraph-online-plugin-scanner/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: "plugin-scanner security audit report (audit version 1)"
version: "unspecified"
type: report
authors:
- name: "hashgraph-online"
date-released: "2026-08-27"
url: "https://skillstore.io/skills/hashgraph-online-plugin-scanner/audits/1"
identifiers:
- type: other
value: "skillstore:hashgraph-online-plugin-scanner:audit:1"
description: "Skillstore immutable audit report identifier"
Skillstore 評分
評分依據 證據置信度: 中 55
架構
85
可維護性
87
內容
65
社群
83
規範符合度
你可以打造什麼
Review a Skill Before Installation
Scan a local skill directory and summarize security findings before adding it to an agent environment.
Audit an MCP Integration
Inspect an MCP server repository for prompt injection, command execution, credential exposure, and suspicious installation behavior.
Validate a Marketplace Submission
Run scan, lint, and verification checks before publishing an agent plugin or skill.
試試這些提示
Scan a Local Skill
Scan [PATH] as an agent skill and summarize the highest severity finding, affected files, and recommended action.
Inspect an MCP Server
Audit the MCP server at [PATH] for prompt injection, secret exposure, unsafe commands, and suspicious installation behavior.
Prepare a Publication Review
Scan, lint, and verify [PATH], then report structural failures, security findings, and publication blockers.
Compare Scan Evidence
Analyze the machine-readable scan of [PATH], group evidence by risk category, and prioritize remediation without claiming the target is safe.
最佳實務
- Scan the narrowest local path that contains the requested target.
- Review cited files and rules before accepting or dismissing each finding.
- Describe clean results as no covered issue detected, not as proof of safety.
避免
- Do not execute target code, install scripts, or lifecycle hooks during scanning.
- Do not read environment files, credential stores, private keys, or unrelated secrets.
- Do not install plugin-scanner or change Python environments without user approval.
常見問題
What can this skill scan?
It guides scans of agent skills, plugins, MCP servers, agent packages, and repositories available through a local path.
Does scanning execute the target?
No. The workflow explicitly avoids running target code, install scripts, lifecycle hooks, and arbitrary target commands.
Is Guard Cloud required?
No. The documented scanner runs locally and does not require Guard Cloud.
Is plugin-scanner included with hol-guard?
No. The skill states that plugin-scanner is a separate Python distribution and command.
Does a clean result prove the target is safe?
No. A clean result means the current scanner found no issue covered by its active rules.
Can the skill install the scanner automatically?
It must ask for approval first. The documented installation uses pipx to isolate the CLI environment.
開發者詳細資訊
授權
Apache-2.0
Skillstore 修訂
r1
版本提示
作者未聲明版本。
引用
40175672d649fa6482bff53a7acf4ecdd6a04a96
維護新鮮度
2026/8/27
使用量
1 次下載 · 0 次瀏覽
檔案結構
📄 SKILL.md
hashgraph-online 的更多技能
查看全部相關技能
查看全部📦
skill-security
71Audit AI Agent Skills Before Installation
由 superagent-ai
Unvetted agent skills can hide prompt injection, credential theft, persistence, or unsafe dependencies. This skill combines offline scanning with guided semantic review.
安全與法遵 查看
📦
vulnerability-scanner
75依據 OWASP 指引審查漏洞風險
由 sickn33
安全審查常因發現項目僅被列出，卻缺少攻擊路徑或業務風險脈絡而有所遺漏。此技能結合 OWASP 指引、本機檢查與優先順序提示，協助進行可執行的安全工作。
安全與法遵 查看
📦
ci-cd-security
75Audit GitHub Actions Workflows
由 superagent-ai
GitHub Actions workflows can expose credentials, repositories, and releases through subtle configuration errors. This skill applies a structured security review and proposes concrete fixes.
安全與法遵 查看
📦
dependency-security
70稽核依賴項安全性並產生 SBOM
由 Doyajin174
依賴項更新可能會引入有漏洞或不受信任的套件。此技能會引導稽核、更安全的更新、SBOM 建立和 CI 檢查。
安全與法遵 查看
📦
Security Researcher
70使用結構化檢查清單稽核安全風險
由 ClementWalter
當團隊仰賴臨時問題時，安全審查可能會遺漏重要控制項。此技能會引導 Claude、Codex 和 Claude Code 針對應用程式、基礎架構、供應鏈、密碼學、智慧合約和 ZK 領域進行結構化稽核。
安全與法遵 查看
📦
dependency-auditor
68稽核依賴項安全性
由 CuriousLearner
依賴項漏洞會造成發布風險與供應鏈曝險。此技能可協助 Claude、Codex 和 Claude Code 審查套件並規劃修復。
安全與法遵 查看