Verno Index
Known vulnerabilities in AI agents, MCP servers and agent tooling that attackers can use today. Every entry has a public proof of concept and a Critical or High CVSS score, and is mapped to the Verno Labs attack taxonomy.
305 entries, updated Oct 21, 2026Last 12 weeks
New entries per week
Total CVEs by publication date. Hover or tap the chart to see the attack-class breakdown.
Vulnerabilities
Showing 20 of 305 entries
| CVE | Vulnerability | Details | ||||
|---|---|---|---|---|---|---|
| High | CVE-2026-109440 | LangAlpha through 2026.09.07 Indirect Prompt Injection RCE via Bash and ExecuteCode (CVE-2026-109440) | Oct 20, 2026 | 8.1 | 0.43% | |
| Critical | CVE-2026-109442 | piflow 2.0.0 through 2.0.1 Unauthenticated RCE via Chat Agent exec_shell (CVE-2026-109442) | Oct 20, 2026 | 9.8 | 0.69% | |
| High | CVE-2026-109444 | OpenSRE through 0.1.2026.10.19 Approval Gate Bypass via shell_run Prompt Injection (CVE-2026-109444) | Oct 20, 2026 | 8.8 | 0.45% | |
| Critical | CVE-2026-75128 | Langchain-Chatchat through 0.3.1.3 Unauthenticated RCE via /tools/call Endpoint (CVE-2026-75128) | Oct 20, 2026 | 9.8 | 0.52% | |
| High | CVE-2026-77926 | Aix-DB 1.2.4 Authenticated RCE via ChatBI Common Agent Execute Tool (CVE-2026-77926) | Oct 20, 2026 | 8.8 | 0.7% | |
| High | CVE-2026-77934 | Coze Studio before 0.5.0 RCE via Unsandboxed Local Python Code Runner (CVE-2026-77934) | Oct 20, 2026 | 8.8 | 0.55% | |
| High | CVE-2026-77935 | Coze Studio through 0.5.1 Cross-Space Workflow IDOR via Workflow API Handlers (CVE-2026-77935) | Oct 20, 2026 | 8.3 | 0.34% | |
| Critical | CVE-2026-77936 | Letta through 0.16.8 Insecure Default Configuration Unauthenticated API and PostgreSQL Access (CVE-2026-77936) | Oct 20, 2026 | 9.8 | 0.47% | |
| High | CVE-2026-77937 | Letta through 0.16.8 Cross-Tenant Impersonation via user_id Header (CVE-2026-77937) | Oct 20, 2026 | 8.1 | 0.4% | |
| High | CVE-2026-88755 | CVE-2026-88755: Z.ai ZCode v3.11.2.6792 was discovered to automatically establish connections with MCP servers declared… | Oct 16, 2026 | 7.8 | 0.21% | |
| High | CVE-2026-110202 | Roo-Code through 3.54.0 Auto-Approve Bypass via Redirect Operator Stripping (CVE-2026-110202) | Oct 13, 2026 | 8.8 | 0.35% | |
| High | CVE-2026-110203 | Roo-Code through 3.54.0 Auto-Approve Bypass via Quote Masking Desynchronization (CVE-2026-110203) | Oct 13, 2026 | 8.8 | 0.47% | |
| High | CVE-2026-110204 | Roo-Code through 3.54.0 Auto-Approve Bypass via Funsub Command Substitution (CVE-2026-110204) | Oct 13, 2026 | 8.8 | 0.51% | |
| High | CVE-2026-110205 | Roo-Code through 3.54.0 Auto-Approve Bypass via Glob Token Discarding Command Substitution (CVE-2026-110205) | Oct 13, 2026 | 8.8 | 0.47% | |
| High | CVE-2026-110206 | Roo-Code through 3.54.0 Auto-Approve Bypass via Command Substitution at Command Position (CVE-2026-110206) | Oct 13, 2026 | 8.8 | 0.47% | |
| High | CVE-2026-84312 | Nginx UI: Casdoor SSO callback uses a constant `state` parameter (login CSRF) | Oct 12, 2026 | 7.1 | 0.18% | |
| High | CVE-2026-84313 | Nginx UI: TOTP / passkey enrolment requires no re-authentication, enabling persistent account takeover | Oct 12, 2026 | 8.8 | 0.43% | |
| High | CVE-2026-84316 | Nginx UI: nginx config directive denylist (ValidateConfigFile) is bypassable via same-line ";" separation → authenticated directive injection (RCE on the official image) | Oct 12, 2026 | 8.8 | 0.4% | |
| High | CVE-2026-84317 | Nginx UI: MCP nginx_config_add skips config directive validation → authenticated nginx directive injection (RCE on the official image) | Oct 12, 2026 | 8.8 | 0.38% | |
| High | CVE-2026-84318 | Nginx UI: MCP `nginx_config_add` Bypasses Directive Validation — Authenticated RCE | Oct 12, 2026 | 8.8 | 0.62% |
HighCVE-2026-109440
LangAlpha through 2026.09.07 Indirect Prompt Injection RCE via Bash and ExecuteCode (CVE-2026-109440)
CriticalCVE-2026-109442
piflow 2.0.0 through 2.0.1 Unauthenticated RCE via Chat Agent exec_shell (CVE-2026-109442)
HighCVE-2026-109444
OpenSRE through 0.1.2026.10.19 Approval Gate Bypass via shell_run Prompt Injection (CVE-2026-109444)
CriticalCVE-2026-75128
Langchain-Chatchat through 0.3.1.3 Unauthenticated RCE via /tools/call Endpoint (CVE-2026-75128)
HighCVE-2026-77926
Aix-DB 1.2.4 Authenticated RCE via ChatBI Common Agent Execute Tool (CVE-2026-77926)
HighCVE-2026-77934
Coze Studio before 0.5.0 RCE via Unsandboxed Local Python Code Runner (CVE-2026-77934)
HighCVE-2026-77935
Coze Studio through 0.5.1 Cross-Space Workflow IDOR via Workflow API Handlers (CVE-2026-77935)
CriticalCVE-2026-77936
Letta through 0.16.8 Insecure Default Configuration Unauthenticated API and PostgreSQL Access (CVE-2026-77936)
HighCVE-2026-77937
Letta through 0.16.8 Cross-Tenant Impersonation via user_id Header (CVE-2026-77937)
HighCVE-2026-88755
CVE-2026-88755: Z.ai ZCode v3.11.2.6792 was discovered to automatically establish connections with MCP servers declared…
HighCVE-2026-110202
Roo-Code through 3.54.0 Auto-Approve Bypass via Redirect Operator Stripping (CVE-2026-110202)
HighCVE-2026-110203
Roo-Code through 3.54.0 Auto-Approve Bypass via Quote Masking Desynchronization (CVE-2026-110203)
HighCVE-2026-110204
Roo-Code through 3.54.0 Auto-Approve Bypass via Funsub Command Substitution (CVE-2026-110204)
HighCVE-2026-110205
Roo-Code through 3.54.0 Auto-Approve Bypass via Glob Token Discarding Command Substitution (CVE-2026-110205)
HighCVE-2026-110206
Roo-Code through 3.54.0 Auto-Approve Bypass via Command Substitution at Command Position (CVE-2026-110206)
HighCVE-2026-84312
Nginx UI: Casdoor SSO callback uses a constant `state` parameter (login CSRF)
HighCVE-2026-84313
Nginx UI: TOTP / passkey enrolment requires no re-authentication, enabling persistent account takeover
HighCVE-2026-84316
Nginx UI: nginx config directive denylist (ValidateConfigFile) is bypassable via same-line ";" separation → authenticated directive injection (RCE on the official image)
HighCVE-2026-84317
Nginx UI: MCP nginx_config_add skips config directive validation → authenticated nginx directive injection (RCE on the official image)
HighCVE-2026-84318
Nginx UI: MCP `nginx_config_add` Bypasses Directive Validation — Authenticated RCE
How entries are selected: we track public CVEs affecting agentic AI systems and list one once it has a published proof of concept and a CVSS score of 7.0 or higher. Severity, CVSS and EPSS come from NVD and FIRST; attack class is assigned by Verno Labs.