Aguara, developed by Garagon, is an open-source static security scanner built primarily with Go. Designed for AI agent skills and Model Context Protocol (MCP) servers, it is distributed as a single signed Go binary. Aguara helps establish trust in repositories and AI agents by performing local, deterministic, and offline scans. It integrates into various development workflows and CI/CD pipelines, offering security checks without external dependencies or telemetry. Its focus is on pre-execution vetting of code and configurations.

What does Aguara connect to?

Aguara connects to AI agent frameworks and MCP clients, functioning as a security scanner for AI agent skills and MCP server configurations. The mcp-aguara component specifically exposes the Aguara engine as an MCP server. Its category is an open-source static security scanner for AI agent and supply-chain trust. Aguara can automatically discover MCP server configurations from 17 popular AI coding assistants and agent frameworks, including Claude Desktop, VS Code, Zed, OpenCode, and OpenClaw clients.

Core features and tools

The mcp-aguara server exposes several callable tools. These are: scan_content, check_mcp_config, list_rules, explain_rule, and discover_mcp. Aguara operates with a local, deterministic, and offline model, requiring no SaaS account, telemetry, or external LLM calls. It utilizes signed releases and signed threat intelligence, with an embedded snapshot in the binary and opt-in signed updates for integrity validation.

Who is it for?

Aguara targets developers, AI agent users, security teams, and CI/CD pipeline operators who need to establish trust in repositories and AI agents. It supports vetting third-party AI agent skills, scanning freshly cloned repositories, and integrating security checks into CI/CD workflows. This tool is ideal for environments prioritizing local, offline security analysis. Skip this if you require a security scanner that relies on cloud-based AI models or external network calls for analysis.

How to get started

No credentials are required for accessing the mcp-aguara server itself. Its primary transport for MCP server functionality is stdio, designed for local, in-process use.

Use cases

• Scanning a freshly cloned repository with "aguara check ." before installing dependencies or executing code to detect malicious packages or insecure configurations.
• Vetting third-party AI agent skills or tools with "aguara scan" before an agent is delegated to use them, to prevent issues like prompt injection or data exfiltration.
• Integrating into CI/CD pipelines using "aguara audit . --ci" to scan CI workflows and agent configurations, with output formats like SARIF for GitHub Code Scanning.

Aguara provides a powerful, offline static security scanning capability for AI agents and MCP servers. Its local execution model and focus on supply-chain trust make it a valuable tool for developers and security teams. The ability to integrate into various AI client frameworks and CI/CD pipelines enhances its utility. Its signed releases and threat intelligence provide confidence in its integrity. Try scanning a freshly cloned repository with "aguara check ." to assess its immediate benefits.