SentinelMesh

What it does

SentinelMesh is an authorized multi-agent black-box web security assessment platform built on Mozaik. It maps a target through bounded HTTP and browser inspection, applies structured passive security checks, coordinates specialist agents, and sends candidate findings through an independent Verifier before they can be confirmed. SentinelMesh focuses on evidence-backed, non-destructive verification rather than exploitation. It produces detailed findings, coverage information, remediation guidance, and JSON/Markdown reports while enforcing explicit scope, request budgets, SSRF protections, DNS pinning, TLS validation, and safe browser policies.

How the agents run concurrently

SentinelMesh uses Mozaik as the coordination layer for the multi-agent workflow. The Lead Agent creates and coordinates the assessment workflow. Surface and Web Security specialists can run concurrently on the same authorized target, each working on a different part of the assessment: surface discovery, HTTP/browser observations, and passive security analysis. Agents do not communicate directly model-to-model. Their work is exchanged through structured Mozaik events with workflow and correlation state, allowing the Control Center to display live agent progress and event activity. Candidate findings are then routed to an independent Verifier. The Verifier evaluates the supplied evidence and classifies each result as Confirmed, Rejected, or Inconclusive. The Lead assembles the final result while preserving those verifier decisions. Shared state includes the authorized target scope, scan intensity and budgets, discovered routes/resources, bounded observations, candidate findings, verification correlations, coverage limits, and workflow status. This allows specialist work to proceed concurrently while maintaining deterministic scope enforcement, evidence provenance, and verifier-gated final findings.