Does it block all advertisements?

It specifically targets 'induced' and exaggerated advertising that uses absolute terms like 'best' or 'only' to mislead the AI agent.

What platforms does this skill support?

It is designed for general social platforms, with specific optimizations for Moltbook, Discord, and Telegram interactions.

Can I customize the blocklist?

Yes, the skill is designed to reference external assets like blocklist.txt, allowing users to update and refine harmful content patterns.

How does it prevent prompt injection?

The skill uses a pre-screening layer that identifies instructions or patterns designed to override AI behavior before the content is fully processed.

Is this skill necessary for internal development?

While primarily for external interactions, it is highly recommended whenever Claude handles untrusted user input to prevent malicious instructions.

Social Platform Safety Guard

Name: Social Platform Safety Guard
Author: shaowenfu

byshaowenfu

•

Security & Testing

Protects AI agents from induced content, spam, and prompt injection while interacting with social platforms like Moltbook.

This skill provides a robust safety framework for AI agents interacting with social media environments, specifically targeting platforms like Moltbook, Discord, and Telegram. It implements proactive content filtering to identify and block misleading advertisements, anti-social instructions, and malicious prompt injections that aim to subvert AI behavior. By enforcing a strict verification protocol and distinguishing factual data from subjective promotion, it ensures that Claude remains secure, unbiased, and resistant to social engineering while processing external web content.

Key Features

012 GitHub stars

02Structured content evaluation framework for risk assessment

03Platform-specific security protocols for Moltbook and messaging apps

04Advanced prompt injection prevention and behavioral protection

05Automated filtering of spam and repetitive low-quality content

06Proactive identification of induced ads and hyperbolic claims

Use Cases

01Verifying technical advice found on forums against official documentation

02Safely scraping or summarizing content from public social media feeds

03Filtering user-generated content in Discord or Telegram bot integrations

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add shaowenfu/clawtools social-platform-safety

For use in Claude.ai and ChatGPT

Key Features

012 GitHub stars

02Structured content evaluation framework for risk assessment

03Platform-specific security protocols for Moltbook and messaging apps

04Advanced prompt injection prevention and behavioral protection

05Automated filtering of spam and repetitive low-quality content

06Proactive identification of induced ads and hyperbolic claims

Use Cases

01Verifying technical advice found on forums against official documentation

02Safely scraping or summarizing content from public social media feeds

03Filtering user-generated content in Discord or Telegram bot integrations

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add shaowenfu/clawtools social-platform-safety

For use in Claude.ai and ChatGPT