Entroly
Open-source context engine that catches AI hallucinations and cuts your token bill 70–95%. The only AI helper that shows its work. Claude · Cursor · Codex,GPT & Custom Providers
What is Entroly?
Entroly is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to open-source context engine that catches ai hallucinations and cuts your token bill 70–95%. the only ai helper that shows its work. claude · cursor · codex,gpt & custom providers
Open-source context engine that catches AI hallucinations and cuts your token bill 70–95%. The only AI helper that shows its work. Claude · Cursor · Codex,GPT & Custom Providers
This server falls under the Data Science & ML category on MCPgee, the world's largest MCP server directory with 33,000+ servers.
Features
- Open-source context engine that catches AI hallucinations an
Use Cases
Maintainer
Works with
Installation
Manual Installation
npx entrolyConfiguration
Configuration Details
claude_desktop_config.json
Performance
Response Metrics
Resource Usage
How to Set Up and Use Entroly
Entroly is an open-source context engine that acts as a transparent proxy between your AI coding tools and LLM providers, compressing context by 70–95% and catching hallucinations via its built-in WITNESS verification system (AUROC 0.844). It wraps 38 popular agents including Claude Code, Cursor, Aider, and Cline, and exposes an MCP server for IDE integration. Developers use it to dramatically cut their API token bills while gaining a live dashboard showing exactly which context was included and which AI claims were suppressed.
Prerequisites
- Python 3.8+ (for the reference runtime) or Node.js 18+ (for the WASM runtime)
- An Anthropic, OpenAI, or other supported LLM provider API key
- Claude Desktop, Cursor, or another MCP-compatible IDE
- pip or npm package manager
Install Entroly
Install the core package with pip for the Python runtime. Add [proxy] for HTTP proxy support, [native] for the faster Rust engine, or [full] for everything. Alternatively install via npm for a Python-free WASM runtime.
pip install entroly[full]Run the auto-setup command
From your repository root, run entroly go. It auto-detects your IDE, wraps your agent with the correct integration method (proxy env-wrap for CLIs, auto-merged mcp.json for MCP IDEs), and opens the dashboard.
cd /your/repo && entroly goOr wrap a specific agent manually
If you prefer to configure a single tool, use entroly wrap followed by the agent name. This integrates Entroly with one of the 38 supported tools without affecting others.
entroly wrap claudeStart the MCP server for IDE integration
For MCP-aware IDEs like Cursor or VS Code, start the MCP server. Entroly will auto-merge its configuration into your mcp.json.
entroly serveConfigure environment variables for proxy mode
For tools that accept custom base URLs (Cline, Continue, JetBrains AI, etc.), point them at the local Entroly proxy. The proxy listens on port 9377 and transparently forwards to upstream providers after compression.
ANTHROPIC_BASE_URL=http://localhost:9377 your-ai-tool
# or for OpenAI-compatible tools:
OPENAI_BASE_URL=http://localhost:9377/v1 your-ai-toolPreview savings on your repository
Run the demo command to see estimated token and cost savings before committing to full integration. This analyzes your repo and shows before/after comparisons.
entroly demoEntroly Examples
Client configuration
Configure Entroly as an MCP server in Claude Desktop. The server command starts the MCP endpoint that handles context compression and hallucination filtering.
{
"mcpServers": {
"entroly": {
"command": "entroly",
"args": ["serve"]
}
}
}Prompts to try
Entroly works transparently — use your AI assistant as normal. These prompts showcase how to verify facts and control context selection.
- "Summarize this 10,000-line log file and verify each finding against the actual content"
- "Refactor the authentication module — use only the files directly relevant to auth"
- "Explain the database schema and confirm every claim against the actual schema file"
- "Show me my token usage breakdown for this session"
- "What context was included in the last request and what was pruned?"Troubleshooting Entroly
entroly go does not detect my IDE or agent
Run `entroly wrap <agent-name>` explicitly with one of the 38 supported agent names (e.g., claude, cursor, aider, cline). Run `entroly wrap --list` to see all supported agents.
Proxy mode causes authentication errors with the upstream provider
Entroly proxies requests to the upstream API using your existing credentials — it does not store them. Ensure ANTHROPIC_API_KEY or OPENAI_API_KEY is set in the same environment where you set ANTHROPIC_BASE_URL or OPENAI_BASE_URL.
WITNESS hallucination detection is too aggressive and suppresses valid responses
The NLI verification model has a threshold that can be tuned. For offline/faster (but less accurate) verification, set ENTROLY_LOCAL_NLI=1. For production, omit this flag to use the default online verification.
Frequently Asked Questions about Entroly
What is Entroly?
Entroly is a Model Context Protocol (MCP) server that open-source context engine that catches ai hallucinations and cuts your token bill 70–95%. the only ai helper that shows its work. claude · cursor · codex,gpt & custom providers It connects AI assistants to external tools and data sources through a standardized interface.
How do I install Entroly?
Follow the installation instructions on the Entroly GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.
Which AI clients work with Entroly?
Entroly works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.
Is Entroly free to use?
Yes, Entroly is open source and available under the Apache-2.0 license. You can use it freely in both personal and commercial projects.
Entroly Alternatives — Similar Data Science & ML Servers
Looking for alternatives to Entroly? Here are other popular data science & ml servers you can use with Claude, Cursor, and VS Code.
Ultrarag
★ 5.6kA Low-Code MCP Framework for Building Complex and Innovative RAG Pipelines
RocketRide
★ 3.1k📇 🏠 - MCP server that exposes RocketRide AI pipelines as t
Aix Db
★ 2.1kAix-DB 基于 LangChain/LangGraph 框架，结合 MCP Skills 多智能体协作架构，实现自然语言到数据洞察的端到端转换。
NeMo Data Designer
★ 1.9k🎨 NeMo Data Designer: Generate high-quality synthetic data from scratch or from seed data.
PaperBanana
★ 1.7kOpen source implementation and extension of Google Research’s PaperBanana for automated academic figures, diagrams, and research visuals, expanded to new domains like slide generation.
MiniMax
★ 1.5kBridges MiniMax AI capabilities to the Model Context Protocol, enabling AI agents to perform image understanding, text-to-image generation, and speech synthesis. It provides a standardized interface for accessing MiniMax's core tools via JSON-RPC.
Browse More Data Science & ML MCP Servers
Explore all data science & ml servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.
Set Up Entroly in Your Editor
Choose your AI client for step-by-step setup instructions.
Quick Config Preview
Add this to your claude_desktop_config.json or .cursor/mcp.json
Ready to use Entroly?
Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.