[AppSec] Prompt Injection & LLM Misuse Scanner

A dedicated scanner for applications that integrate LLMs. It analyzes code calling LLM SDKs (Anthropic, OpenAI, Bedrock and Vertex auto-detected) and flags OWASP LLM Top 10 patterns: user input concatenated into prompts, missing guardrails or system prompt enforcement, indirect prompt injection through unvalidated tool output, unsafe tool use (file, exec or network access without allowlists), missing rate or cost limits, and secrets embedded in prompts. Findings are reported alongside your SAST, SCA and IaC results, with the same prioritization and remediation workflow.