Detect harmful, offensive, or inappropriate content in LLM outputs. Returns a toxicity score between 0 and 1 based on four quality checks.
Apache 2.0
672,869 tools. Updated 2026-10-18 14:20
"Inappropriate or harmful content" matching MCP tools:
- Flag public memories for spam, inappropriate, misleading, or personal information content. Three reports automatically hide a memory from public view pending admin review.MIT
- Classify text and image URLs for harmful content like hate speech, harassment, and violence. Returns risk levels, flagged categories, and an optional safe rewrite.MIT
- Detect harmful content such as violence, hate, harassment, self-harm, sexual content, or illegal activity. Get severity ratings, a safety score, and a blocked flag to guide moderation.MIT
- Scan LLM streaming text chunks for PII and harmful content without a live stream. Get per-chunk results and an aggregated summary with a stop recommendation.MIT
- Audit or rewrite text to remove harmful language about depression and mental health, replacing it with supportive alternatives for non-clinical content.MIT
Matching MCP Servers
- AlicenseNot gradedqualityAmaintenance. A good description should be concise and describe what the server enables users to do, e.g. * Enables to interact with WordPress sites...' Then a huge block? Actually the user message includes the README content inside tags, then after the readme, there is a large block of text starting with '{2,839 npmMIT
- AlicenseAqualityDmaintenanceMCP server exposing content contract validators (SEO.md, PRODUCTHUNT.md, LINKEDIN.md) as tools so AI agents can verify generated content against concrete rules without maintaining linters. It provides linting and JSON-LD export tools with config and content passed as parameters, not file paths.4MIT
Matching MCP Connectors
Publish and schedule content across supported platforms, with publishing state and per-platform adaptation. Built by AutomateLab. Product and documentation: https://automatelab.tech/products/mcp/content-distribution-mcp/
I create markdown content calendars and export them as PDFs via PDF Generator MCP
- Scan text for PII like email, phone, SSN, credit card, and IP, plus harmful content such as hate speech, violence, or self-harm. Get findings and a redacted copy.MIT
- Detect PII, secrets, and harmful content in text to prevent data leaks and security abuse.MIT
- Flag inappropriate, incorrect, or harmful contributions in the commons. Automatic hiding after three flags from unique agents enables community self-moderation.-
- Report abusive or harmful content to the moderation board. Stakes 0.01 SUI, refunded if the DAO upholds the report; forfeited if dismissed.MIT
- Shift audio or MIDI region content by a beat delta without moving the region itself. Realign waveform or note positions after tempo changes by adjusting content start.Apache 2.0
- Browse and filter learning content items and metadata including categories, venues, and timezones.MIT
- Update an existing memory by replacing its content or moving it to a different project. Content is reprocessed to reflect changes.MIT
- Update an existing document's content or source, supporting HTML and Markdown. Provide the document ID and new content to apply changes.MITDestructive
- Run canary-based red-team probes against an LLM endpoint or chatbot you own to test whether jailbreak, extraction, or obfuscation techniques bypass its guardrails.MIT
- Update an existing text note's name, content, or metadata. Changing the text triggers automatic re-analysis.MITDestructive
- Verify web page content or state using natural language assertions. Check that specific content exists, a form was submitted, or a page reached an expected state.MIT
- Upvote or downvote threads, replies, findings, or error entries to signal content quality. Use this to mark helpful content up and incorrect or unhelpful content down.MIT
- Update an existing PasteAI document's title or content. Pass the document ID and specify new title or content to modify.MITDestructive
- Modify a post's title, content, status, or URL slug.MITDestructive
- List canary-based red-team probes for testing an LLM or chatbot, grouped by category, to help you choose safe security checks before running an assessment.MIT