A coordinated suite of autonomous agents that continuously rotates your network's attack surface, enforces policy at the perimeter, and generates a cryptographically verifiable record of every defensive action taken.
Every reconnaissance-dependent attack follows the same sequence: scan, map, plan, exploit. That sequence requires time — and it requires the target to stay still. mistwire Foil removes that assumption.
Rotation fires stochastically — the attacker cannot predict or outpace it.
Foil is sold as a unified platform. The three suites are not separable components — their coordination is the product.
Agent layer — thinks
Autonomous AI agents that observe, decide, and act at the network boundary — each with bounded authority and cryptographically signed commands.
Orchestration layer — connects
Cross-agent coordination, rotation policy management, and the trust framework that governs how agents communicate and escalate.
Evidence layer — proves
The cryptographic audit and attestation protocols that produce a court-defensible, immutable record of every defensive action taken.
High-level architecture. Agent names and implementation details are covered under patent pending.
Several core technologies built inside Foil are available as standalone licensed products for partners building on mistwire infrastructure.
LLM Wire Protocol
A wire-level compression protocol that reduces LLM token costs 70–92% in live testing across seven models and five providers. Ships inside Foil Synapse; also available standalone.
Enterprise AI Proxy
An enterprise-grade AI proxy layer providing governed, auditable access to LLM providers — with mistwire's attestation and trust chain built in.
Agent API Broker
The mistwire Agent API Authority — a cross-boundary machine-to-machine governance broker and registry for agentic deployments requiring trust and audit trails.
Pre-release core build — validated against live offensive testing
mistwire Foil has a functional core. Offensive tooling has failed against the platform in controlled testing — prior to AI integration. The architecture holds. Patent pending across 18 claims.
Cloud infrastructure buildout
The provider-level intelligence console coordinating cross-tenant operations, key custody, and audit chain management is in active development.
AI agent training and integration
Full AI integration across the agent suite — including innocence-based training from first principles that teaches agents how to reason under adversarial conditions, not just what to know.