mcp-security-project/awesome-agentic-mcp-security
mcp-security-project/awesome-agentic-mcp-security has 26 GitHub stars and 39 forks. GitHot measured 0 stars in 24 hours. Last checked 2026-11-15 19:34 UTC. Last push 2026-11-10.
What you can do
- Model Context Protocol Specification - Official MCP specification.
- Security Best Practices - Confused deputy, token passthrough, SSRF, session hijacking, and scope minimization.
- OWASP MCP Top 10 - Community risk taxonomy for MCP-enabled systems.
- MCP Azure Security Guide - Microsoft guide mapping the OWASP MCP Top 10 to concrete Azure mitigations.
Measured star growth
- in 24h
- 0
- in 7 days
- Update delayed
- in 15 days
- Update delayed
- in 1 month
- Update delayed
Net star changes over overlapping periods, not daily totals. Missing measurements are left blank. Stars measure attention, not software quality.
About
Awesome Agentic MCP Security List
LearningREADME.md
OriginalAwesome Agentic MCP Security Awesome
Curated links on agentic AI and Model Context Protocol security: threats, research, tooling, labs, and learning resources.
Maintained and curated by Vandana Verma Sehgal.
Each topic page is a standalone reference with tables, summaries, and outbound links.
This list treats MCP as a new security boundary and curates 400+ resources across blogs, papers, talks, scanners, integrations, labs, CVEs, and videos. Each topic page is a standalone reference with tables, summaries, and outbound links.
Contents
- Highlights — start here
- Project overview
- Blogs, whitepapers and research
- Conference talks and briefings
- Free trainings and courses
- GitHub repos and discovery
- Podcasts and webinars
- Security tooling
- MCP security servers and integrations
- CVEs and advisories
- Vulnerable environments and labs
- YouTube and video library
- Code of Conduct
Highlights — start here
If you only have a few minutes, these are widely-cited starting points across the list.
Foundations
- Model Context Protocol Specification - Official MCP specification.
- Security Best Practices - Confused deputy, token passthrough, SSRF, session hijacking, and scope minimization.
- OWASP MCP Top 10 - Community risk taxonomy for MCP-enabled systems.
- MCP Azure Security Guide - Microsoft guide mapping the OWASP MCP Top 10 to concrete Azure mitigations.
- MCP Security Cheat Sheet - OWASP cheat sheet covering tool trust, validation, and supply-chain hygiene.
Foundational research and incidents
- Tool Poisoning Attacks - Invariant Labs' early notification on malicious instructions hidden in tool descriptions.
- Jumping the line: How MCP servers can attack you before you ever use them - Trail of Bits.
- MCP Servers: The New Security Nightmare - Equixly's widely-cited empirical vulnerability rates in popular MCP servers.
- Building Secure MCP Servers: A Developer's Guide - Snyk developer-focused guide.
Hands-on
- Damn Vulnerable MCP Server (DVMCP) - CTF-style lab for prompt injection, tool poisoning, rug pulls, and shadowing.
- MCP Inspector RCE (CVE-2025-49596) - Critical advisory in MCP developer tooling.
- Snyk Agent Scan - CLI scanner for prompt injection, tool poisoning, and toxic flows in agent and MCP configurations.
Project overview
Terminology, MCP as a trust boundary, threat framing, audience notes, and how the topic lists fit together with pointers to the MCP specification and OWASP MCP Top 10.
Key risk areas covered: Data leakage, unauthorized actions, prompt injection chains, tool poisoning, supply chain compromise, shadow MCP servers, and weak monitoring/auditability.
Blogs, whitepapers and research
Browse all blogs and research →
70+ blogs, whitepapers, standards, and academic papers with short summaries for theme-based reading. Includes Microsoft, Anthropic, Invariant Labs, Trail of Bits, Snyk, Wiz, OX Security, JFrog, CyberArk, Equixly, Red Hat, Stack Overflow, OWASP, and Cloud Security Alliance. Plus a chronological "Dated incidents" subsection covering AgentFlayer, mcp-remote RCE (CVE-2025-6514), MCPoison, Asana cross-tenant exposure, Cato CTRL PoCs, ANSI deception, and other notable incidents.
Selected reading:
- Toxic Flow Analysis - Invariant Labs.
- Insecure credential storage plagues MCP - Trail of Bits.
- The Cursor Agentic Jira MCP Attack Explained with Toxic Flow Analysis - Liran Tal.
- Is that allowed? Authentication and authorization in Model Context Protocol - Stack Overflow Blog.
Conference talks and briefings
30+ talks and briefings where MCP security is a primary focus, with speakers, venues, links to recordings or slides where public, and relevance to exploits, tool trust, or governance. Spans Black Hat USA, Black Hat Europe, Black Hat Asia, RSAC, OWASP, BSides, SecTor, and regional chapter events.
Recent highlights: BlueHat Redmond 2026: Securing the Future of AI: Securing MCP with Defense in Depth Patterns. CoSAI at RSAC 2026: Securing MCP: Mitigating New Threats in Agentic AI Deployments (Sarah Novotny, Jason Clinton). Black Hat Europe 2025: MCP Unchained: Compromising The AI Agent Ecosystem Via Its "Universal Connector". Black Hat USA 2025: From Prompts to Pwns: Exploiting and Securing AI Agents (Becca Lynch, Rich Harang). OWASP GenAI @ RSAC 2026: Securing MCP: OWASP Best Practices - A Practical Guide (Idan Habler, Joshua Beck, Tomer Elias). BSides Seattle 2026: MCP LFI in 60 Minutes (or Your Money Back) (Kurt Boberg).
Free trainings and courses
Free or low-barrier structured learning for MCP fundamentals and security-aware integration, with notes on format, duration, and sequencing. Organized by skill level (beginner, intermediate, advanced) and split between MCP fundamentals, MCP security, and company-sponsored trainings.
Selected:
- Official MCP Documentation: Getting Started - Model Context Protocol project.
- MCP for Beginners - Microsoft's open-source curriculum across multiple languages.
- Model Context Protocol Course - Hugging Face structured course.
GitHub repos and discovery
50+ official SDKs, reference servers, registries, discovery hubs, scanners, frameworks, and "awesome lists" with brief rationale and security-oriented caveats per row.
For official SDKs, reference servers, scanners, and discovery hubs, see the full catalog in mcp_github_repos.md.
Podcasts and webinars
Browse all podcasts and webinars →
Audio and live sessions on MCP security and related agentic risk, with metadata for discovery and executive-friendly context.
Selected:
- OWASP MCP Top 10: 2026 Security Framework and MCP Security Certification - Practical DevSecOps.
- #249: The MCP Security Risks You Can't Afford to Ignore - Tech Lead Journal / Ariel Shiftan (MCPTotal).
- Vibe Hacking: AI-Driven C2 Operations with MCP - SANS.
- AI-Driven DevSecOps Part 1: Securing Coding Agents with MCP - SANS.
- Lessons for Security Leaders From the MCP AI Supply Chain Crisis - OX Security.
Security tooling
50+ scanners, monitors, policy and gateway controls, secrets and dependency checks, and related open-source utilities with deployment notes and safety warnings where relevant. Organized by category: MCP scanners, runtime monitoring, policy engines, secrets and dependency scanners, red team tools, and open-source observability.
See mcp_security_tools.md for
- MCP scanners,
- runtime monitoring/policy options, and
- hardening utilities (with deployment notes and safety warnings).
MCP security servers and integrations
60+ MCP servers that wrap AppSec and SAST tools, dependency and OSV data, reverse engineering and mobile stacks, threat intel APIs, cloud and identity platforms, SIEM platforms, and trust or reputation layers (including agent-market style integrations).
Categories: Categories include
- AppSec/SAST/dependencies,
- reverse engineering/mobile,
- threat intelligence/OSINT,
- cloud/identity/zero-trust,
- SIEM/security analytics, and
- hosting/gateways/transport helpers.
CVEs and advisories
Structured catalogs of published CVEs affecting MCP servers, clients, SDKs, and related tooling.
Notable:
See mcp_cve.md for mcp-cve-project (571 CVE rows synced; upstream README states 570 coverage) and vulnerablemcp.
Vulnerable environments and labs
Intentionally vulnerable MCP setups, goats, CTF-style labs, and PoCs with setup hints, themes, and isolation reminders.
Selected:
See mcp_vulnerable_environments_and_labs.md for vulnerable labs and PoCs, plus isolation and safety notes.
YouTube and video library
Video primers and security-focused sessions with channel, speaker, date, duration, links, and suggested learning outcomes. Organized into general primers, security videos, demo and PoC content, and CVE / exploit walkthroughs.
Selected:
- The OWASP MCP Top 10 — Every Risk Ranked and Explained - StackOps AI.
- Securing Model Context Protocol (MCP) with Vandana Verma (OWASP/Snyk) - Long-form session on production MCP security.
- First Look - OWASP MCP Top 10 - 2025 - Vandana Verma.
- MCP Security: Why Your AI Assistant Is an Insider Threat - Liran Tal, Snyk.
Contributors
We'd like to thank the following contributors for their valuable input:
- Rohit G (for sharing the mcp repo) and
- Liran Tal (for sharing the mcp repo).
- Randall Degges (for sharing the incredible feedback and suggesting the changes).
Code of Conduct
Community participation follows the Code of Conduct.