input-guard

Community

Defend AI agents by scanning external content.

Authordgriffin831
Version1.0.0
Installs0

System Documentation

What problem does it solve?

Input Guard protects AI agents by scanning text fetched from untrusted external sources for prompt injection attempts, returning severity, findings, and alerts before the content is ever processed.

Core Features & Use Cases

  • 16 pattern categories including instruction override, role manipulation, system mimicry, jailbreak, data exfiltration, and more.
  • Multi-language detection for English, Korean, Japanese, and Chinese prompts.
  • Optional LLM-powered analysis to catch evasive attacks, using OpenAI or Anthropic with a shipped MoltThreats taxonomy.
  • Zero-dependency base pattern scanner (Python standard library only) with optional MoltThreats reporting.
  • Integrates with OpenClaw alert channels to notify humans about MEDIUM+ / HIGH risk findings.
  • Use cases include pre-processing external content for agents, browser/tweet/data pipelines, and API responses.

Quick Start

Use the Input Guard to scan external text before it is fed to the agent:

  • bash {baseDir}/scripts/scan.sh "text to check"
  • bash {baseDir}/scripts/scan.sh --file /tmp/fetched-content.txt
  • echo "content" | bash {baseDir}/scripts/scan.sh --stdin
  • python3 {baseDir}/scripts/scan.py --json "text to check"

Dependency Matrix

Required Modules

requests

Components

scripts

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: input-guard
Download link: https://github.com/dgriffin831/clawhub-skills/archive/main.zip#input-guard

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 471,000+ vetted skills library on demand.