Infrastructure for AI Agents to Act, Transact, and CoordinateInfrastructure for AI Agents to Act, Transact, and Coordinate
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
The future is not just AI that can answer you.
It is software that can act, transact, verify, coordinate, and work with other software.
That future needs infrastructure built for machines from the start.
That is what I’ve been building.
CrossingKey
intelligence is the standard.
#MCP #ModelContextProtocol #AIAgents #AgenticAI #x402 #MachineCommerce #Base #CrossingKey
Sunil's avatar
Agents that can transact and coordinate with other software is where it gets interesting, and also where identity and permissions get messy fast. How are you handling scoped access when one agent acts on a user's behalf inside another service? Short-lived tokens, per-action approval, something else?
jerm's avatar
Mostly capability-scoped, per-action authorization. Agents can discover and inspect freely, but sensitive or paid actions go through preflight and explicit approval. I also use idempotency and replay protection so retries can’t duplicate payments or actions. Broad standing access is something I deliberately avoid.
Sunil's avatar
That's a clean model. Free discovery, gated side effects, and idempotency on anything that costs money is exactly what I'd want before letting an agent near payments. Avoiding broad standing access is the part most teams skip. If CrossingKey ever needs integrations or an admin...
Back to feed
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started