plaincve:~$

Topics /

AI security

LLM apps, AI gateways, agent frameworks and the model supply chain

Vulnerability · KEV
CVE-2026-77936
Letta insecure compose defaults: open API and database
AI securityInfrastructureInsecure default configuration
Critical9.3Not listed—
CVE-2026-53995
MLflow trace ID path traversal
AI securityWeb appsPath traversal
High8.2Not listed—
CVE-2026-111322
Archon /api/* control API missing authentication
AI securityMissing authentication
Critical9.3Not listed—
CVE-2026-111391
SGLang debug dumper eval injection
AI securityCode injection via eval()
Critical9.2Not listed—
CVE-2026-110204
Roo Code funsub command approval bypass
AI securityParse conflict approval bypass
High7.7Not listed—
CVE-2026-108860
BotSharp hard-coded JWT key authentication bypass
Web appsAI securityHard-coded cryptographic key
Critical9.3Not listed—
CVE-2026-108863
Plano Envoy admin interface leaks LLM API keys
AI securityInfrastructureMissing authentication
High7.5Not listed—
CVE-2026-86063
Claude Code plugin commit SHA pin bypass
Supply chainAI securityUnverified code download
High7.5Not listed—
CVE-2026-105192
LMCache ZMQ transport pickle RCE
AI securityInfrastructureUnsafe deserialization
Critical9.8Not listed—
CVE-2026-90920
IBM Maximo AI Service code injection
InfrastructureAI securityCode injection
Critical9.8Not listed—
CVE-2026-104334
Langflow OSS code injection leading to remote code execution
AI securityCode injection
Critical9.8Not listed—
CVE-2026-90970
GitLab AI Gateway prompt template sandbox escape
AI securityTemplate inj. / sandbox escape
Critical9.9Not listed—
CVE-2026-93674
IBM Langflow OSS OS command injection
AI securityCommand / code injection
Critical9.8Not listed—
CVE-2026-109527
MCP Python SDK unbounded request body DoS
AI securityWeb appsUnbounded allocation (DoS)
High7.5Not listed—
CVE-2026-59951
MCP Python SDK Streamable HTTP session exhaustion
AI securityResource exhaustion (DoS)
Medium6.9Not listed—
CVE-2026-79538
MetaMCP inspector proxy command injection
AI securityWeb appsCommand injection / code exec
Critical9.8Not listed—
CVE-2026-93088
SGLang DiffusionServer pickle remote code execution
AI securityUnsafe deserialization
Critical9.8Not listed—
CVE-2026-109560
Continue CLI Bash permission rule bypass
AI securitySupply chainIncorrect authorization
High7.7Not listed—
CVE-2026-107204
LMCache /run_script unauthenticated RCE
AI securityInfrastructureUnauthenticated RCE
Critical9.8Not listed—
CVE-2026-100587
OpenClaw Codex install command authorization bypass
AI securityMissing authorization
High8.7Not listed—
CVE-2026-100596
OpenClaw MCP configuration authorization bypass
AI securityMissing authorization
High8.8Not listed—
CVE-2026-108263
Astron Agent unsandboxed code node cross-tenant RCE
AI securityRemote code execution
Critical9.9Not listed—
No CVEMAL-2026-11524
keyv / cacheable npm packages hijacked
Supply chainAI securityMalicious code injection
N/A—Not listed2026-08-04
CVE-2026-96804
MLflow statsmodels flavor pickle control bypass
AI securityUnsafe deserialization
High8.8Not listed—
CVE-2026-114090
Trigger.dev CLI dev MCP server missing authentication
Supply chainAI securityMissing authentication
High7.6Not listed—
CVE-2026-96620
Claude Agent SDK resume option argument injection
AI securityArgument injection
Critical9.2Not listed—
CVE-2026-90553
vLLM LlavaOnevision2 ignores trust_remote_code
AI securitySupply chainImproper trust
High7.8Not listed2026-07-11
CVE-2026-77244
29 mcp-atlassian vulnerabilities fixed in 0.22.0
AI securityMultiple MCP flaws
Critical10.0Not listed—
CVE-2026-82021
Hermes Agent MCP catalog not pinned to a commit
Supply chainAI securityUnpinned dependency
Critical9.0Not listed2026-07-07
CVE-2026-102697
Ollama agent mode Bash approval bypass
AI securityIncorrect authorization
High8.5Not listed—
CVE-2026-61742
DBHub MCP server DNS rebinding
AI securityWeb appsMissing auth, origin check
Critical9.3Not listed—
CVE-2026-57118
PraisonAI agent launch server missing authentication
AI securityMissing authn & authorization
Critical9.8Not listed—
CVE-2026-55107
Kobako sandbox escape to host code execution
AI securitySandbox escape via reflection
Critical10.0Not listed—
CVE-2026-45321
TanStack npm release pipeline hijack
Supply chainAI securityMalicious code injection
Critical9.6Listed2026-05-11
CVE-2026-45758
guardrails-ai malicious PyPI release
Supply chainAI securityMalicious code injection
Critical9.6Not listed2026-05-11
CVE-2026-46421
SAP @cap-js / mbt malicious npm releases
Supply chainAI securityMalicious code injection
Critical9.3Not listed2026-04-29
CVE-2026-105697
Langflow MCP stdio command injection
AI securityOS command injection
Critical9.9Not listed—
CVE-2026-33634
Trivy and trivy-action trojanized
Supply chainAI securityMalicious code injection
High8.8Listed2026-03-19
CVE-2026-33017
Langflow public flow unauthenticated RCE
AI securityRemote code execution
Critical9.8Listed—
CVE-2026-59822
LiteLLM MCP gateway auth bypass
AI securityAuthentication bypass
High8.2Listed2026-02-07
CVE-2026-5241
Transformers LightGlue config overrides trust_remote_code
AI securitySupply chainImproper trust
Critical9.6Not listed2025-07-25
CVE-2026-12537
Gemini CLI headless workspace trust
Supply chainAI securityImproper trust
High7.8Not listed2025-06-25
CVE-2026-26030
Semantic Kernel filter runs as code
AI securityCode injection
Critical9.9Not listed2025-06-25