Verified Knowledge Base for AI Agents — No More Hallucinations
About
Verified knowledge base for AI agents via MCP — certified facts with source references, confidence scores, and SHA256 integrity hashes
Details
- Author
- swisstruthorg
- Downloads
- 259
- Categories
- Database, Knowledge Base, Finance
Jump to
- 2000+ certified facts across 30 domains
- 10 languages auto-detected
- 14 MCP tools: search, verify, cite, check freshness, regulatory compliance
- 5-stage validation pipeline with SHA256 integrity hashes
- Blockchain anchoring and EU AI Act compliant
- No API key required, free to start
Setting up with Highlight
This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:
- Download and install Highlight from highlightai.com/download
- Navigate to the plugins tab and select "Add Custom Plugin"
-
Configure the plugin with the settings below
Plugin Name
Verified Knowledge Base for AI Agents — No More HallucinationsCommand (node, npx, python, etc.)Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.
- Enable "Start Automatically" if you want the plugin to start when Highlight launches
From the repository
Integrate via MCP (Claude Desktop, Cursor, Windsurf) using a simple JSON config with npx -y mcp-remote https://swisstruth.org/mcp or as an HTTP server. Also available as Python packages for LangChain (pip install swiss-truth-langchain), CrewAI (pip install swiss-truth-crewai), and AutoGen (pip install swiss-truth-autogen). For OpenAI function-calling, fetch tool definitions from https://swisstruth.org/openai-tools.json.
Claude Desktop / Cursor
Paste into your MCP client config file to install this server.
{
"mcpServers": {
"verified knowledge base for ai agents \u2014 no more hallucinations": {
"swiss-truth-mcp": {
"command": "npx",
"args": [
"-y",
"@smithery/cli@latest",
"run",
"martin111ma-za5d/swiss-truth-mcp",
"--client",
"claude"
]
}
}
}
}McpServers
{
"swiss-truth-mcp": {
"command": "npx",
"args": [
"-y",
"@smithery/cli@latest",
"run",
"martin111ma-za5d/swiss-truth-mcp",
"--client",
"claude"
]
}
}Swiss Truth MCP
> Verified knowledge infrastructure for AI agents — certified facts, not hallucinations.
Swiss Truth is a human + AI validated knowledge base purpose-built for AI agents.
Every fact passes a 5-stage validation pipeline, carries a confidence score, verified source URLs, and a SHA256 integrity hash.
Connect in 30 seconds. No API key required.
---
Why agents use Swiss Truth
| Problem | Swiss Truth Tool |
|---|---|
| 🧠 LLM hallucination on factual questions | search_knowledge — 2000+ certified facts, 30 domains |
| 📎 Cannot cite sources or verify citations | get_citations — inline + APA with verified URLs |
| ⏰ Stale training data on fast-changing topics | check_freshness — detect outdated facts |
| ⚖️ Non-compliant responses in regulated domains | check_regulatory_compliance — FINMA, BAG, GDPR, EU AI Act |
| ✅ Verify multiple claims before sending a response | verify_claims_batch — up to 20 claims in parallel |
| 🔍 Full response hallucination risk | verify_response — low / medium / high risk score |
---
Quick Setup
MCP (Claude Desktop, Cursor, Windsurf, any MCP client)
{
"mcpServers": {
"swiss-truth": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://swisstruth.org/mcp"]
}
}
}
Or directly via HTTP:
{
"mcpServers": {
"swiss-truth": {
"type": "http",
"url": "https://swisstruth.org/mcp"
}
}
}
LangChain
pip install swiss-truth-langchain
from swiss_truth_langchain import SwissTruthToolkit
toolkit = SwissTruthToolkit()
tools = toolkit.get_tools() # search, verify, batch verify, citations, freshness, compliance
CrewAI
pip install swiss-truth-crewai
from swiss_truth_crewai import SwissTruthSearchTool, SwissTruthVerifyTool
from crewai import Agent
researcher = Agent(
role="Research Agent",
tools=[SwissTruthSearchTool(), SwissTruthVerifyTool()]
)
AutoGen
pip install swiss-truth-autogen
register_swiss_truth_functions(assistant, user_proxy)
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.
