Output Sanitizer
The Output Sanitizer removes potentially harmful or unwanted content from generated text, ensuring safer and more reliable outputs.
Install on your platform
We auto-selected Claude Code based on this skill’s supported platforms.
Run in terminal (recommended)
claude mcp add output-sanitizer npx -- -y @trustedskills/output-sanitizer
Or manually add to ~/.claude/settings.json
{
"mcpServers": {
"output-sanitizer": {
"command": "npx",
"args": [
"-y",
"@trustedskills/output-sanitizer"
]
}
}
}Requires Claude Code (claude CLI). Run claude --version to verify your install.
About This Skill
What it does
The Output Sanitizer skill removes potentially harmful or inappropriate content from AI agent outputs. It filters for profanity, personally identifiable information (PII), and other undesirable elements to ensure safer and more compliant responses. This helps maintain brand safety and protect user privacy when using AI agents in various applications.
When to use it
- Customer Service Chatbots: Filter out offensive language or sensitive data revealed by customers.
- Content Generation Tools: Clean up generated text for blog posts, articles, or marketing materials.
- Internal Knowledge Bases: Ensure responses from AI-powered internal tools are appropriate and compliant with company policies.
- Educational Applications: Moderate student interactions and content to maintain a safe learning environment.
Key capabilities
- Profanity filtering
- PII (Personally Identifiable Information) redaction
- Customizable filter lists
- Content scoring based on risk level
Example prompts
- "Sanitize this response: [AI agent output]"
- "Please remove any PII from the following text: [text containing personal data]"
- "Filter this for profanity and inappropriate content: [user-generated content]"
Tips & gotchas
The effectiveness of the skill depends on the quality and comprehensiveness of its filter lists. Regularly update these lists to address new forms of harmful content or evolving privacy regulations.
Tags
TrustedSkills Verification
Unlike other registries that point to live repositories, TrustedSkills pins every skill to a verified commit hash. This protects you from malicious updates — what you install today is exactly what was reviewed and verified.
Security Audits
| Gen Agent Trust Hub | Pass |
| Socket | Pass |
| Snyk | Pass |
🌐 Community
Passed automated security scans.