hoop
One gateway in front of every protocol. Same policy across MCP, LLMs, databases and containers. Wire-level enforcement at under 5ms.
Hoop is an open-source MCP server that functions as a security sidecar, providing runtime control and governance for AI agents accessing databases, APIs, and other resources. It sits transparently between agents and their data sources, enforcing wire-level policies like data masking, guardrails against destructive operations, and session analysis—all under 5ms latency. Agents require no SDK changes; they simply point to the sidecar's port instead of the direct resource. The single-binary, MIT-licensed tool uses one YAML config file to apply consistent security policies across PostgreSQL, SQL Server, HTTP, and MCP protocols.
Key Features
Use Cases
- 01Preventing AI agents from executing destructive database operations like table drops or mass deletions
- 02Redacting personally identifiable information (emails, phone numbers) from query results before agents process them
- 03Enforcing consistent security policies across multiple AI agents accessing production databases
- 04Reviewing and approving high-risk agent actions through human escalation before execution
- 05Auditing all agent-database interactions with wire-level inspection and event logging
- 06Protecting multi-protocol environments where agents use MCP servers, direct database connections, and HTTP APIs
Related MCP Servers
View moremarkitdown
Python tool for converting files and office documents to Markdown.
firecrawl
The context API to search, scrape, and interact with the web at scale. 🔥
prompts.chat
f.k.a. Awesome ChatGPT Prompts. Share, discover, and collect prompts from the community. Free and open source — self-host for your organization with complete privacy.
langflow
Langflow is a powerful tool for building and deploying AI-powered agents and workflows.
hoop — FAQ
What is the Hoop MCP server?+
Hoop is an open-source security sidecar that provides runtime controls for AI agents accessing databases and APIs. It enforces data masking, guardrails, and risk analysis at the wire protocol level without requiring changes to agent code or prompts.
How do I install Hoop?+
On macOS, install via Homebrew with 'brew tap hoophq/brew' and 'brew install hoop'. For other platforms, download the binary from the GitHub releases page. Start the sidecar with 'hoop start sidecar' (uses default config) or 'hoop start sidecar --config config.yaml' for custom policies.
Which AI clients and agents work with Hoop?+
Hoop works with any AI agent or MCP client that connects to PostgreSQL, SQL Server, or HTTP resources. It operates transparently at the wire protocol level, so agents simply point to the sidecar's listening port instead of the direct database or API endpoint.
Do I need API keys or special credentials to use Hoop?+
No API keys are required for basic operation (data masking and guardrails). The optional Session Analyzer feature can integrate with AI models for risk scoring, which may require model access credentials. The sidecar itself runs locally with just a YAML config file.
Is Hoop free to use?+
Yes, Hoop is open source and released under the MIT license, making it free for commercial and personal use. The control plane for managing sidecar fleets is also included in the repository.
How do I configure data masking and guardrails?+
Create a YAML config file defining mask rules (entity types to redact), policy rules (operations to block), and listeners (protocols and ports). Point the sidecar to this config with '--config config.yaml', and your agents connect to the sidecar's listen port instead of the upstream resource.
How do I install hoop?+
Open the source repository on GitHub and follow its README. hoop is a mcp server — MCP Agents Market links you directly to the official repo.
Is hoop free?+
hoop is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.