Sharing a project for anyone interested in AI agents and APIs: SAIHM (Sovereign AI Horizontal Memory), an open, AI-agent-agnostic protocol — it works with any agent regardless of framework or model — that gives an agent persistent, encrypted memory surviving across sessions. Memories are shareable: you can grant another agent or person access to specific ones, and a user can cryptographically erase any of it on request.
Why it might interest this community: SAIHM speaks plain HTTP/JSON-RPC, so every operation (remember, recall, forget, status, share) can be explored straight from Postman — no SDK required. Each call is an ordinary request/response you can inspect, save to a collection, and replay.
The on-chain pieces run on COTI V2, and the client is Apache 2.0 if you want to read or fork it. You decide what gets stored, shared, and forgotten — the protocol keeps nothing you didn’t ask it to.
Would genuinely like to know what this community would want in a public collection — it could be shaped around how you’d actually test the API.
Join SAIHM at https://saihm.coti.global