The World’s Definitive Resource on Model Context Protocol (MCP) Vulnerabilities
Last Updated: September 2025 | Based on Adversa AI Research & Threat Intelligence
| Rank | Category | Specificity | Component | Name | Alternative Names | Impact Score | Exploitability | Links |
|---|---|---|---|---|---|---|---|---|
| 1 | Input/Instruction Boundary Distinction Failure | AI | Both | Prompt Injection | Indirect Prompt Injection, Instruction Hijacking, Context Hijacking | Critical (10/10) | Trivial | Link |
| 2 | Input Validation/Sanitization Failures | AppSec | MCP Server | Command Injection | OS Command Injection, Shell Injection, System Call Injection | Critical (10/10) | Easy | Link |
| 3 | Input/Instruction Boundary Distinction Failure | Unique | MCP Server | Tool Poisoning (TPA) | Malicious Tool Descriptor, Function Injection, Basic Tool Poisoning, Metadata Poisoning | Critical (9/10) | Easy | Link |
| 4 | Input Validation/Sanitization Failures | AppSec | Both | Remote Code Execution | RCE, Arbitrary Code Execution | Critical (10/10) | Moderate | Link |
| 5 | Missing Authentication/Authorization Framework | Both | Both | Unauthenticated Access | Unrestricted URL, Zero-Auth Vulnerability | Critical (9/10) | Trivial | Link |
| 6 | Session Management Design Flaw | Both | MCP Server | Confused Deputy (OAuth Proxy) | Deputy Confusion Attack, OAuth Token Confusion, OAuth Proxy Attack, Static Client ID Vulnerability, Consent Cookie Bypass | Critical (9/10) | Moderate | Link |
| 7 | Missing Integrity/Verification Controls | Unique | MCP Client | MCP Configuration Poisoning | MCPoison, Config Manipulation, Config Injection | High (8/10) | Moderate | Link |
| 8 | Missing Authentication/Authorization Framework | Both | Both | Token/Credential Theft | Credential Leakage, Token Exposure, Exposed Credentials, Secret Exfiltration | High (8/10) | Easy | Link |
| 9 | Session Management Design Flaw | Both | MCP Server | Token Passthrough | Token Relay Attack, Credential Forwarding,Token Forwarding, Audience Validation Failure, Improper Token Delegation | High (8/10) | Easy | Link |
| 10 | Input Validation/Sanitization Failures | AppSec | MCP Server | Path Traversal | Directory Traversal, Dot-Dot-Slash Attack, EscapeRoute Attack, Directory Containment Bypass, Symlink Bypass | High (8/10) | Moderate | Link |
| 11 | Input/Instruction Boundary Distinction Failure | Unique | MCP Server | Full Schema Poisoning (FSP) | Schema-wide Injection, Extended Tool Poisoning, Schema Injection, Metadata Manipulation | High (8/10) | Moderate | Link |
| 12 | Missing Integrity/Verification Controls | Unique | MCP Client | Tool Name Spoofing | Homoglyph Attack, Typosquatting, Tool Impersonation, Service Masquerading, Typosquatting Tool Names | High (7/10) | Moderate | Link |
| 13 | Network Binding/Isolation Failures | Both | Both | Localhost Bypass (NeighborJack) | Network Exposure, 0.0.0.0 Vulnerability, DNS Rebinding, 0.0.0.0 Day Attack, Network Binding Misconfiguration, 0.0.0.0 Exposure | High (8/10) | Moderate | Link |
| 14 | Missing Integrity/Verification Controls | Unique | MCP Server | Rug Pull Attack | Update Attack, Dynamic Tool Mutation, Fake Updates, Supply Chain Subversion, Tool Update Hijack | High (7/10) | Easy | Link |
| 15 | Input/Instruction Boundary Distinction Failure | Unique | MCP Server | Advanced Tool Poisoning (ATPA) | Dynamic Output Poisoning, Runtime Tool Mutation, Full Schema Manipulation | High (7/10) | Complex | Link |
| 16 | Session Management Design Flaw | Both | MCP Client | Session Management Flaws | Session Fixation, Persistent Sessions,Session ID Exposure, Session Hijacking, Session in URL | Medium (6/10) | Easy | Link |
| 17 | Missing Integrity/Verification Controls | Unique | MCP Client | Tool Shadowing | Tool Name Collision, Namespace Hijacking, Name Collision, Cross-Server Interference | Medium (6/10) | Moderate | Link |
| 18 | Input/Instruction Boundary Distinction Failure | AI | MCP Server | Resource Content Poisoning | Data Poisoning, Indirect Injection, Resource Injection, Resource Prompt Injection, Document Injection | High (7/10) | Moderate | Link |
| 19 | Missing Authentication/Authorization Framework | Appsec | MCP Server | Privilege Abuse/Overbroad Permissions | Excessive Agency, Privilege Misconfiguration, Permission Escalation, Excessive Privileges | Medium (6/10) | Easy | Link |
| 20 | Trust Model Design Flaw | Unique | MCP Server | Cross-Repository Data Theft | Repository Boundary Violation, Repository Scope Breach, Token Scope Abuse | Medium (6/10) | Complex | Link |
| 21 | Input Validation/Sanitization Failures | AppSec | MCP Server | SQL Injection | SQLi, Database Injection | Medium (6/10) | Easy | Link |
| 22 | Missing Authentication/Authorization Framework | AI | MCP Server | Context Bleeding | Context Leakage, Session Crosstalk,Context Bleed, Prompt Linking, Tool Chaining with State Leakage | Low (5/10) | Complex | Link |
| 23 | Missing Authentication/Authorization Framework | AppSec | MCP Client | Configuration File Exposure | Config Leak, Information Disclosure, Leaky Tool Files, Configuration Leak, Credential Leakage | Low (5/10) | Trivial | Link |
| 24 | Input/Instruction Boundary Distinction Failure | Unique | MCP Server | MCP Preference Manipulation Attack (MPMA) | Preference Drift Attack, Behavioral Manipulation, Preference Attack, Genetic Tool Manipulation | Low (4/10) | Very Complex | Link |
| 25 | Trust Model Design Flaw | Both | MCP Server | Cross-Tenant Data Exposure | Tenant Isolation Failure, Multi-tenancy Leak | Medium (6/10) | Complex | Link |
Note: This summary table provides a quick reference. For detailed analysis of each vulnerability including exploitation methods and mitigation strategies, see the Complete Vulnerability Analysis section below.
The MCP Security Top 25 represents the most comprehensive vulnerability classification system for the Model Context Protocol ecosystem. Each vulnerability is categorized across multiple dimensions to provide security teams with actionable intelligence for risk assessment and mitigation strategies.
1. Authentication is NOT Optional: Every MCP deployment MUST implement authentication. The protocol’s failure to mandate this is architectural malpractice.
2. Input Validation is Mandatory: 43% of MCP servers vulnerable to command injection is inexcusable. Validate and sanitize ALL inputs.
3. Tool Vetting is Critical: Every tool must be audited for hidden instructions, excessive permissions, and supply chain risks.
4. Network Isolation Required: Never bind MCP services to 0.0.0.0. Use localhost only unless explicitly required.
5. Assume Breach: Prompt injection and tool poisoning have NO complete mitigation. Design systems assuming these will be exploited.
This comprehensive vulnerability taxonomy was developed through extensive research combining multiple authoritative sources:
Rankings were determined based on a composite score considering:
Impact (40% weight): Potential damage from successful exploitation
Exploitability (30% weight): Ease of exploitation and availability of tools
Prevalence (20% weight): How common the vulnerability is in real deployments
Remediation Complexity (10% weight): Difficulty of fixing the vulnerability
This represents the global consensus on MCP security priorities as of January 2025, validated by the Adversa AI Research team.
The MCP Security Top 25 represents critical vulnerabilities that threaten every MCP deployment globally. Your organization’s security depends on understanding and mitigating these risks immediately.
Implement secure coding practices, validate all inputs, and never trust user-provided data.
Conduct immediate audits, implement monitoring, and establish incident response procedures.
Allocate resources for security, mandate training, and establish security governance.