XTRIKE AI SUPPLY CHAIN & AI-BOM

XTRIKE AI Inventory & AI-BOM Security Hub

Inventory, verify, and monitor AI model weights, agentic RAG pipelines, prompt injection vectors, and MCP server endpoints.

Discovered AI Assets

1,420 Models

MCP Server Endpoints

38 Active Tools

Prompt Injection Risks

2 Critical

AI-BOM Components & Model Lineage

CycloneDX AI-BOM 1.6

AI-M-01xtrike-llm/Core-8B-Instruct

LOW RISK

Source: XTRIKE Model Registry

Format: Safetensors (GGUF)

Injection Risk: Guarded (XTRIKE PromptShield v3)

Vulnerabilities: None (SHA-256 Verified)

AI-M-02xtrike-rag-agent-v2

CRITICAL RISK

Source: Internal Repo / Agent Chain

Format: Python Agentic Pipeline

Injection Risk: Direct Indirect Injection in User Prompt

Vulnerabilities: CVE-2024-Agent-Eval (Untrusted Code Exec)

AI-M-03xtrike-embeddings/MiniLM-L6-v2

LOW RISK

Source: XTRIKE Embeddings Hub

Format: ONNX Runtime

Injection Risk: N/A (Embedding Only)

Vulnerabilities: Clean Model Weights

AI-M-04mcp-filesystem-server-v1

HIGH RISK

Source: Model Context Protocol (MCP)

Format: JSON-RPC Stdio Server

Injection Risk: Directory Traversal via Tool Arguments

Vulnerabilities: Path Escape in fs_read tool