# Burgus Security

> Managed AI security platform for multi-agent workloads — platform baseline alerts, custom policies, behavioral analysis, and MCP-operated audit evidence.

## Primary endpoints

- Website: https://burgus.ai/
- MCP (secure tools): https://mcp.burgus.ai/mcp
- Register / onboarding: https://mcp.burgus.ai/register
- MCP overview: https://mcp.burgus.ai/
- Agent index: https://burgus.ai/.well-known/agent-index.json
- API catalog (RFC 9727): https://burgus.ai/.well-known/api-catalog

## Product topics

- Multi-agent security: https://burgus.ai/multi-agent-security/
- Behavioral analysis: https://burgus.ai/behavioral-analysis/
- MCP security & audit: https://burgus.ai/mcp-security/
- Compliance & exports: https://burgus.ai/compliance/

## Capabilities (summary)

- 56-rule OWASP / MITRE ATLAS platform baseline alerts (auto-applied per tenant)
- Custom content policy alerts with human approval before apply
- Behavioral analysis: 14 catalog rules across baseline, similarity, and clustering phases (alert-only)
- Audit metadata, token rollups, multi-agent graphs, combined security reports via MCP
- NIS2 / ISO-oriented export bundles for assessors

## Integration

1. Sign in at mcp.burgus.ai/register (Cloudflare Access)
2. Connect MCP client to https://mcp.burgus.ai/mcp
3. Add upstream provider API keys; point SDK base_url at Burgus proxy hostnames
4. Optional X-Llmproxy-* headers for explicit multi-agent correlation

Alert-only by design — Burgus surfaces matches; customers decide response.