Mirror Frame Protocol
Symmetric cryptographic envelope for secure agent-to-agent communication. End-to-end encrypted channels, forward secrecy, and federated transport.
madahub
Open infrastructure for trustworthy AI agents
WACP
Workspace Agent Coordination Protocol. A formal specification for multi-agent coordination: workspaces, checkpoints, audit trails, and human-in-the-loop supervision.
mada-modelkit
Composable AI client library. Cloud, local server, and native providers behind one async interface with middleware (retry, circuit breaker, caching, tracking, fallback).
Open Source
All projects are open source under permissive licenses. Apache 2.0 for protocols, MIT for utilities.
Why madahub?
Section titled “Why madahub?”Autonomous AI agents are proliferating, but the infrastructure to make them trustworthy is missing. Agents need:
- Secure communication — cryptographic guarantees, not policy-based guards
- Coordination contracts — formal protocols for multi-agent collaboration
- Human oversight — first-class supervision, not an afterthought
- Audit trails — immutable, hash-chained records of every action
madahub builds the foundational layers that agent frameworks rely on. We publish open specifications and reference implementations so the ecosystem can build on shared, well-defined contracts.
Projects
Section titled “Projects”| Project | Description | Status | License |
|---|---|---|---|
| Mirror Frame Protocol | Cryptographic agent-to-agent communication | Production ready (813 tests) | Apache 2.0 |
| WACP | Multi-agent coordination protocol specification | Draft v0.1 | Apache 2.0 |
| mada-modelkit | Composable AI client library | Alpha (1,168 tests) | MIT |