Security scanner for AI agents, MCP servers, and agent skills to detect prompt injections and vulnerabilities.
agent-scan is easy to set up with strong trust signals. Check agent compatibility and use-case fit before adding it to your workflow.
Repository setup guidance
gh repo view snyk/agent-scan --webOpen the official repository or website.
Check the README for package manager, auth, and platform requirements.
Try it in a small test task inside your agent workflow.
Agent Scan helps you find security problems in AI agents and their components. It scans for prompt injections, malware, and other threats in agent skills and MCP servers. You can use it to keep an inventory of all installed agent components.
Agent Scan is a security scanner developed by Snyk that helps developers and security professionals discover and scan AI agent components on their machine for prompt injections and vulnerabilities. It supports scanning of various agents including Claude, Cursor, Windsurf, Gemini CLI, Amp, Amazon Q, and others. The tool auto-discovers MCP configurations, agent tools, and skills, and detects over 15 distinct security risks such as prompt injection, tool poisoning, tool shadowing, toxic flows, malware payloads, untrusted content, credential handling issues, and hardcoded secrets. It provides detailed reports and supports ignoring skill analysis with the --no-skills flag. The scanner requires explicit user consent before executing MCP server commands, and it is recommended to run scans inside a sandbox when evaluating untrusted configurations.
Strong trust signals; still review the README and permissions before production use.
Last commit was about 52 days ago.
2534 GitHub stars indicate community interest.
23 open issues signal maintenance load.
Apache-2.0 license detected.
Scanning AI agent configurations for prompt injection vulnerabilities
Auditing MCP servers for security risks before deployment
Inventorying all installed agent components across a development machine
Detecting malware payloads hidden in natural language agent skills
Integrating security scanning into CI/CD pipelines for agent-based applications
Scanning MCP configurations executes commands defined in them; run in a sandbox for untrusted configs.
Requires explicit user consent before starting each stdio MCP server during interactive scans.
2,534
Stars
229
Forks
23
Issues
Apache-2.0
License
Run large language models locally with a simple CLI. Supports Llama, Mistral, Gemma, and 100+ models with one command.
Universal memory layer for AI agents that enables personalized, context-aware interactions.
An open-source framework to turn HTML, CSS, and animations into deterministic MP4 videos.
Security scanner for AI agents, MCP servers, and agent skills to detect prompt injections and vulnerabilities.
Security engineers auditing AI agent ecosystems, Developers building or integrating AI agents and MCP servers, DevOps professionals managing agent infrastructure
The setup section provides repository-level starting guidance, not a guarantee of an independently verified installation. Check the official README and release notes.
Scanning MCP configurations executes commands defined in them; run in a sandbox for untrusted configs. Requires explicit user consent before starting each stdio MCP server during interactive scans.
Similar or complementary options to evaluate include Ollama, mem0, hyperframes.
Jun 6, 2026
2 security/trust notes recorded.
Setup difficulty is 2/5.