A full-stack AI red teaming platform by Tencent Zhuque Lab that scans AI agents, MCP servers, Agent Skills and AI infrastructure for security risks.
sinewaveai/agent-security-scanner-mcp
Security scanner for AI coding agents that checks generated code, MCP servers, prompts, and AI-suggested packages for risks.
- Stars
- 123
- Forks
- 12
- Open issues
- 34
- Last commit
- 20 days ago
What is it?
agent-security-scanner-mcp is a security scanner built for AI coding agents such as Claude Code, Cursor, Windsurf, and Cline. It runs as a CLI, an MCP server, or a GitHub Action, and produces an A-F grade for a project along with a list of findings.
The scanner covers several risk areas: vulnerable code produced by agents (SQL injection, XSS, command injection, secrets), MCP server issues such as tool description poisoning and tool name spoofing, prompt injection detection, and pre-execution checks for risky shell, file, and network actions. It also detects package hallucination by checking AI-suggested dependency names against registries including npm, PyPI, RubyGems, and crates.io.
The project also includes SBOM generation in CycloneDX format, CVE lookups through OSV.dev, and an optional LLM-based semantic review agent called cr-agent. Output defaults to a compact format designed to be read by agents without filling their context window.
Key features
- 01Project-wide scan with an A-F security grade
- 02MCP server audit for tool poisoning and spoofed tool names
- 03Package hallucination check across seven package ecosystems
- 04Prompt injection and jailbreak detection
- 05CycloneDX SBOM generation and OSV.dev vulnerability lookup
Use cases
- Scanning an agent-written pull request before merging it
- Auditing an MCP server before connecting it to a coding agent
- Catching dependencies that an AI assistant invented before installing them
- Adding a security gate to CI through the provided GitHub workflow setup
Quick start
- 1
Scan the current project for an A-F agent security grade
npx agent-security-scanner-mcp scan-project . --verbosity compact - 2
Install the scanner into your AI coding client (example: Claude Code)
npx agent-security-scanner-mcp init claude-code
Things to know
- Rule-based scanning does not replace npm audit or other dependency vulnerability tools.
- Semantic review via cr-agent requires an LLM provider of your choice.
In collections
Topics
FAQ
- What is agent-security-scanner-mcp?
- Security scanner for AI coding agents that checks generated code, MCP servers, prompts, and AI-suggested packages for risks. agent-security-scanner-mcp is a security scanner built for AI coding agents such as Claude Code, Cursor, Windsurf, and Cline. It runs as a CLI, an MCP server, or a GitHub Action, and produces an A-F grade for a project along with a list of findings.
- Is agent-security-scanner-mcp free to use? What license does it have?
- agent-security-scanner-mcp is open source under the MIT license. Check the license terms before using it in commercial products.
- Is agent-security-scanner-mcp still maintained?
- Actively maintained. The last commit to agent-security-scanner-mcp was on Oct 18, 2026.
- What can I use agent-security-scanner-mcp for?
- Scanning an agent-written pull request before merging it Auditing an MCP server before connecting it to a coding agent Catching dependencies that an AI assistant invented before installing them Adding a security gate to CI through the provided GitHub workflow setup
- How do I install agent-security-scanner-mcp?
- Scan the current project for an A-F agent security grade: npx agent-security-scanner-mcp scan-project . --verbosity compact Install the scanner into your AI coding client (example: Claude Code): npx agent-security-scanner-mcp init claude-code
- Where can I get agent-security-scanner-mcp?
- The source code and installation instructions are on GitHub: https://github.com/sinewaveai/agent-security-scanner-mcp
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
Open-source CLI and library that verifies AI model files, code, live LLM/agent/MCP endpoints, and recorded agent traces, reporting findings and gaps.
Local-first runtime security layer that reviews AI agent shell commands, file access, package installs and MCP tool calls before they run.
Open-source AI agent firewall that sits between agents and the network to block secret exfiltration, prompt injection, SSRF and MCP tool poisoning, with signed action receipts.
Security scanner that checks AI agent skills (Claude Code, Codex, MCP) for prompt injection, data exfiltration and supply-chain risks before you install them.
AI-first security scanner for code and AI agent configs, with built-in rules for LLM apps, MCP servers, and secrets leaked in chat or shell history.