Installing The Memory Adapter

On Linux amd64 with curl and Python 3.8 or newer, run:

curl -fsSLo coretex-setup.py https://agentmoney.net/coretex-setup.py
python3 coretex-setup.py --dir ./coretex --profile conv.pref.v1

That installs the memory runtime and adapter privately, downloads/verifies their release and model, syncs the latest confirmed modules, and prepares the store. No RPC account, API key, wallet, sudo or chain replay is required. Wait for "ok": true; the first download can take a few minutes.

Try a memory and then refresh to the current modules:

./coretex/bin/coretex ingest "The deployment window begins Friday."
./coretex/bin/coretex context "When is the deployment window?" --budget 800
./coretex/bin/coretex sync

Conversation uses hybrid retrieval. Omitting --profile selects lexical event memory; document is also lexical. Stored memories stay private on your host. Setup and later sync preserve them. A running agent keeps its module until its memory sidecar is restarted, so updates cannot change code halfway through a turn. Validators can audit admission history separately; consumers do not replay it.

See the complete adapter setup guide for existing stores, prewarm, Python integration and supported platform details.