# CodeThreat AI Hub - AI Component Analysis

> Security intelligence hub for MCP servers and Agent Skills.
> By CodeThreat. https://hub.codethreat.com

## About
CodeThreat AI Hub is a public security intelligence platform that scans AI components
(MCP servers, Agent Skills) from 10+ registries and publishes
risk profiles with vulnerability findings mapped to MITRE ATLAS and
OWASP Top 10 for Agentic Applications.

## Pages
- [Home](https://hub.codethreat.com/)
- [Search](https://hub.codethreat.com/search)
- [API Documentation](https://hub.codethreat.com/api-docs)
- [Skill Universe](https://hub.codethreat.com/skill-universe)
- [Repo Risk](https://hub.codethreat.com/repo-risk)

## API
- GET /api/v1/artifacts — List artifacts with filters (type, severity, registry, risk level)
- GET /api/v1/artifacts/:id — Artifact detail with risk profile and findings
- GET /api/v1/artifacts/:id/cyclonedx — CycloneDX AI/ML-BOM export
- GET /api/v1/findings — List findings with filters
- GET /api/v1/findings/:id — Finding detail with ATLAS mapping
- GET /api/v1/search — Search across artifacts, findings, and registries
- GET /api/v1/stats — Ecosystem statistics and health metrics
- GET /api/v1/registries — List monitored registries
- GET /api/v1/repo/risk?repo=owner/repo — GitHub repository risk analysis
- GET /api/v1/repo/badge?repo=owner/repo — SVG badge for repository risk

## Standards
- MITRE ATLAS (AML.Txxxx technique mapping)
- OWASP Top 10 for Agentic Applications 2026
- OWASP Top 10 for LLM Applications 2025
- CycloneDX AI/ML-BOM
- Package URL (purl)
- MCPSecBench methodology

## Entity Types
- mcp_server — MCP (Model Context Protocol) servers with tool definitions
- skill — Agent Skills (Cursor Skills, Copilot Extensions, etc.)

## Data Model
- Artifacts: name, canonical_name, artifact_type, description, repository_url, keywords, server_json, skill_md_content, agent_card_json, remote_urls, popularity_metrics
- Findings: severity (CRITICAL/HIGH/MEDIUM/LOW/INFO), title, description, evidence, category, aitech_code (ATLAS), aitech_name, rule_id, remediation, confidence, analyzer
- Risk Profiles: risk_score (0-100), risk_level, finding_counts (JSONB)
- Scan Jobs: scanner_type, status, analyzers, exit_code
- Data Sources: source_type, name, base_url, registry_name