Questions?
Answers.

MAREF (Multi-Agent Recursive Evolution Framework) is an open-source agent governance operating system. It sits between your AI agents and the tools they call, providing 8 layers of defense, formal verification (TLA+), cryptographic audit trails, recursive self-evolution, and Chinese national cryptography (SM2/SM3/SM4-GCM). Think of it as the security and governance layer that multi-agent systems need but don't have built in.

Those frameworks help you <em>build</em> multi-agent systems. MAREF helps you <em>govern</em> them. They solve orchestration — who talks to whom. MAREF solves safety — what agents are allowed to do. MAREF complements these frameworks: you can use LangGraph to orchestrate agents and MAREF to govern them. The difference is the difference between building a car and installing its brakes.

No. MAREF is a governance layer, not a model runtime. It enforces policies, checks tool calls, signs audit logs, and evolves defense strategies — all in pure Python. It runs on a $5 VPS just as well as on a workstation. If you're running LLM agents, those need GPUs; MAREF does not.

Yes. MAREF is released under the Apache 2.0 license. The full source code is available on GitHub at github.com/maref-org/maref. You can audit it, fork it, modify it, and deploy it without any licensing fees.

The AIP Pioneer Program is MAREF's early-adopter initiative. Organizations that join get priority support, direct access to the engineering team, influence over the roadmap, and validated compliance documentation for regulatory audits. It's designed for teams deploying multi-agent systems in regulated industries.

MAREF implements full GB/T 32918 compliance: SM2 for digital signatures and key exchange (256-bit ECC), SM3 for cryptographic hashing (256-bit), and SM4-GCM for authenticated symmetric encryption (128-bit block cipher with Galois/Counter Mode). All implementations are pure Python, zero native dependencies, and fully auditable.

Lyapunov stability analysis mathematically proves that MAREF's governance engine gets safer over time — it doesn't just change, it converges toward a provable minimum error state. Most security systems degrade or oscillate. MAREF's error rate monotonically decreases. It's the difference between 'we hope this works' and 'we can prove it works.'

The Trust Engine v2 evaluates five factors per interaction: recency, consistency, alignment with goals, outcome quality, and adversarial resistance. Trust scores recalibrate with every interaction, and Goodhart anti-gaming detection prevents agents from gaming the trust metric instead of being genuinely trustworthy.

Yes. MAREF is framework-agnostic. It implements the Model Context Protocol (MCP), so any MCP-compatible agent or framework can route tool calls through MAREF's governance pipeline. Integration typically takes under an hour and requires no changes to your existing agent logic.

The 4-level decision tree routes only ~3% of decisions to human review. False positives (blocking legitimate actions) are flagged by the human reviewer, fed back into the evolution engine, and the system learns from the correction. Over 200 rounds of adversarial evolution, the false positive rate dropped by 82%.