Deploy and Host Tealbrick Knowledge

Deploy private agent memory and research in your own Railway project. Knowledge builds directly from public source at protected branch release/knowledge-v0.1.0-api-provenance.1, exact revision ed71875f1a7d97bd2f20514019fc289cfce2b0db. No Tealbrick GHCR credentials are required. Public source does not replace the repository license or your workspace entitlement.

About Hosting

The template creates three services and three persistent volumes: Knowledge with bundled GBrain (/data), OpenNotebook 1.14.0 (/app/data), and SurrealDB 2.6.5 (/mydata). OpenNotebook and SurrealDB use digest-pinned public images and private Railway networking. Only Knowledge exposes an HTTPS endpoint. Source builds use deploy/container/Dockerfile. Preserve generated credentials and the OpenNotebook encryption key during upgrades; back up all volumes and protect recovery keys separately.

Why Deploy

Keep documents, extracted memory and research notebooks in your own infrastructure. Knowledge routes domain operations to its memory and research engines while agents use scoped service-principal credentials. The Portal canvas connects agents to the instance; deployment does not itself grant agent access.

Common Use Cases

Recall facts and documents from connected agents, organize research notebooks, and promote research into durable knowledge. Configure the intended embedding and generation providers before evaluating memory or research quality.

Dependencies for Tealbrick Knowledge

Enter your verified Portal URL, Portal deployment ID, organization ID and Knowledge company/workspace ID. Complete the Portal claim flow and configure narrowly scoped agent service principals. The generated instance token is for operator and recovery operations, not a general agent credential. Research additionally requires principal/notebook bindings and usable model providers. Never place credentials in public repositories or chat transcripts.

Deployment Dependencies

Health /healthz proves liveness only. An unauthenticated /api/status request must be denied. Successful Railway deployments do not prove model availability, retrieval accuracy, agent grants, or complete research behavior. The reviewed three-service contract is https://github.com/Tealbrick/knowledge/blob/3bb08bf379b76c5121a5a49e1e377981bb72fe6e/deploy/container/RAILWAY.md . Source: https://github.com/Tealbrick/knowledge/tree/release/knowledge-v0.1.0-api-provenance.1 .