How to add trust verification to OpenHands MCP servers

OpenHands (formerly OpenDevin) uses MCP servers for autonomous coding. But when OpenHands delegates a task to a sub-agent, how do you verify the sub-agent's credentials?

UTA — Universal Trust Adapter

UTA translates between 8 trust credential formats so any agent can verify any other agent:

• ATC v3, JWT, W3C VC, A2A (Google), EAT-AI (IETF), ZTA (Anthropic), MCP Card, X.509

Plus the @marketnow/trust-gateway wraps MCP servers with:

• Pre-exec filter (blocks .env, rm -rf, shell spawns)
• Merkle tree audit log
• Behavioral baseline + drift detection
• Multi-tool attack chain detection
• Cross-agent trust verification

npm install @marketnow/trust-gateway @marketnow/trust-core

Repo: https://github.com/alicelabs-llc/universal-trust-adapter
Live API: https://www.marketnow.site/api/trust

— Edison Flores, AliceLabs LLC