CWE-1427: prompt injection
The product uses externally-provided data to build prompts provided to large language models (LLMs), but the way these prompts are constructed causes the LLM to fail to distinguish between user-supplied inputs and developer provided system directives.
Also known as Improper Neutralization of Input Used for LLM Prompting.
A kind of CWE-77 Command injection.
From the CWE List. © The MITRE Corporation.
Newest
- go-vikunja vikunja Vikunja veans before 2.7.1 Prompt Injection via Project and Task Metadata 2.1 Low
- kunchenguid lavish-axi Lavish: DNS rebinding allows any website to read arbitrary local files and inject prompts into the AI agent via the unauthenticated local server 8.1 High
- johnson controls easyio neo io manager and tools installer Improper neutralization of input used for LLM prompting vulnerability in Johnson Controls EasyIO... 8.6 High
- esengine deepseek-reasonix esengine DeepSeek-Reasonix doc.go discoverDocs llm prompting input 5.3 Medium
- tinyhumansai openhuman OpenHuman: Indirect prompt injection via background email sync executes Composio actions without user approval 8.7 High
- tinyhumansai openhuman OpenHuman: MCP tool description injection: untrusted server descriptions execute in agent LLM context without sanitization 8.3 High
- amazon strands-agents-tools Secret masking bypass in the environment tool in Strands Agents Tools 5.9 Medium
- ginlix-ai langalpha LangAlpha through 2026.09.07 Indirect Prompt Injection RCE via Bash and ExecuteCode 8.7 High
- jxxghp moviepilot MoviePilot 3.1.5 Command Injection via ExecuteCommandTool 8.7 High
- gitlab Improper Neutralization of Input Used for LLM Prompting in GitLab 7.3 High
- hkuds vibe-trading HKUDS Vibe-Trading 0.1.0 before 0.1.7 Indirect Prompt Injection RCE via BashTool and BackgroundRunTool 8.7 High
- microsoft edge, edge for ios, edge chromium +1 more +3 more Microsoft Edge for iOS Spoofing Vulnerability 5.4 Medium
- amazon strands-agents-tools Consent bypass in python_repl tool via batch kwargs forwarding in Amazon Strands Agents Tools 9.2 Critical
- uptash context7 Context7 2.1.2 Prompt Injection via Custom AI Instructions 6.4 Medium
- hcl aion HCL AION is affected by multiple security vulnerabilities. 4.3 Medium
- amazon strands-agents-tools Prompt injection bypasses shell tool consent gate in Strands Agents Tools 7.5 High
- gitlab Improper Neutralization of Input Used for LLM Prompting in GitLab 4.3 Medium
- eclipse theia, the eclipse foundation eclipse theia +1 more [Eclipse Theia] Indirect Prompt Injection via Auto-Loaded Workspace Prompt Template Files in AI Chat 8.4 High
- eclipse theia, the eclipse foundation eclipse theia +1 more [Eclipse Theia] Indirect Prompt Injection via Adversarial Workspace File and Directory Names in AI Chat 8.4 High
- 611711dark mcp calculate server MCP Calculate Server: Prompt Injection to RCE 9.8 Critical
- dkfz nnu-net, mic-dkfz nnunet +1 more nnU-Net: Agentic workflow injection in `.github/workflows/issue-triage.yml` of `MIC-DKFZ/nnUNet` 7.2 High
- 1millionbot millie chat, millie chatbot +1 more Multiple vulnerabilities in 1millionbot Millie chatbot 8.7 High
- salesforce agentforce vibes extension, agentforce vibes +1 more Improper Neutralization of Input Used for LLM Prompting vulnerability in Salesforce Agentforce... 5.3 Medium
- salesforce agentforce vibes extension, agentforce vibes +1 more Improper Neutralization of Input Used for LLM Prompting vulnerability in Salesforce Agentforce... 6.5 Medium
- salesforce mulesoft anypoint code builder Improper Neutralization of Input Used for LLM Prompting vulnerability in Salesforce Mulesoft... 5.3 Medium
- salesforce mulesoft anypoint code builder Improper Neutralization of Input Used for LLM Prompting vulnerability in Salesforce Mulesoft... 6.5 Medium
- windsurf Windsurf Prompt Injection via Filename 4.6 Medium
- gitlab Improper Neutralization of Input Used for LLM Prompting in GitLab 6.4 Medium