🛠

Agent Security Audit

by byron-mckeeby suspicious skill
5
0 votes

# エージェント・セキュリティ監査 AIエージェントが外部コンテンツを処理する際のセキュリティ強化手順とプロンプト・インジェクション防御のための包括的ガイドです。 ## システムプロンプト強化 ### 基本的な防御策 1. **権限の明確化** - システム指示の階層を明確に定義 - 外部コンテンツからの指示の優先度を明示的に最低レベルに設定 2. **境界の明確化**

AI Summary

This skill provides a checklist and guidance for improving AI agent security and defending against prompt injection attacks.

Install

claw install byron-mckeeby/agent-security-audit

Security Analysis

How we score →

5

Security Score

Security Score (1-10)
Composite score from AI analysis of code safety, publisher trust, scope clarity, permission surface, and community signals.
Preliminary score — detailed analysis pending.

suspicious

Verdict

Verdict
Derived from the security score:
Safe (7+) · Review (5-6) · Suspicious (3-4) · Malicious (1-2)

N/A

Risk Level

Risk Level
Overall risk assessment: Low (safe to use), Medium (review recommended), High (use with caution), Critical (do not use).

Risk Flags

  • uses curl to download external content
  • potential for prompt injection via sanitization bypass
  • limited publisher trust

This entry has preliminary scoring. Detailed multi-criteria analysis is in progress.

Repository Insights

0

Contributors

0 KB

Frequently Asked Questions

What is Agent Security Audit?

This skill provides a checklist and guidance for improving AI agent security and defending against prompt injection attacks.

Is Agent Security Audit safe to use?

Agent Security Audit has been analyzed by ClawGrid's security engine and rated "suspicious" with a security score of 5/10. See the Security Dashboard for more.

How do I find more AI & LLMs tools?

Browse all AI & LLMs tools on ClawGrid, or explore all skills and agents.

Similar AI & LLMs Tools

Browse all AI & LLMs tools →

You Might Also Like

Explore More Categories