AgentGuard: Open-Source AI Agent Security Scanner

[dockfixlabs]

What is AgentGuard?

AgentGuard is an open-source static analysis tool that scans AI agent source code for security vulnerabilities. It detects the OWASP ASI Top 10.

Why does this exist?

AI agents are deployed everywhere - coding assistants, customer support, trading bots, autonomous systems. These agents have access to tools, APIs, filesystems, and network resources. Nobody is scanning their code for agent-specific vulnerabilities.

Existing tools (Bandit, Semgrep, CodeQL) scan for traditional vulnerabilities. AgentGuard scans for agent-specific attack vectors: prompt injection, tool abuse, data exfiltration, credential exposure, excessive agency, supply chain risks, and more.

Install

pip install dfx-agentguard
agentguard ./src

MCP Server Mode (Claude Code / Cursor)

{"command": "agentguard", "args": ["mcp"]}

Ecosystem

• AgentGuard - Core scanner
• MCP Scanner - MCP server security scanner
• VS Code Extension - Inline diagnostics
• GitHub App - Automated PR reviews
• Benchmark - 27 vulnerable code samples

Roadmap

• v0.3.0: Semantic analysis, GitHub Action, pre-commit hook
• v0.4.0: REST API, web dashboard, custom rules
• v1.0.0: Enterprise SSO, audit logs, on-premise

Full roadmap: https://github.com/dockfixlabs/agentguard/blob/main/ROADMAP.md

Contributing

We welcome bug reports, feature requests, and pull requests. See CONTRIBUTING.md.

Feedback

What agent security challenges are you facing? What rules would you like to see?