activegraph-gbrain-bridge

This repository is a compatibility bridge between two Python libraries called ActiveGraph and GBrain. Both relate to building AI agent systems, and the bridge connects them while keeping their responsibilities separate.

GBrain handles durable, persistent knowledge: structured notes and documents stored in files, retrieval indexes, typed relationships between concepts, facts, timelines, and data ingestion jobs. Think of it as a long-term memory and knowledge store for an AI agent. ActiveGraph handles the runtime side: recording what an agent actually did, logging events, tracking cause-and-effect chains between steps, replaying past runs, and enforcing policy rules before allowing certain actions. Together, the two libraries give an AI agent both a place to store what it knows and a record of how it used that knowledge.

This bridge package, installable as a Python library, provides a typed client interface for reading from GBrain, querying its schema, and proposing changes back to it in a dry-run mode. A dry-run means the proposed change is submitted for review rather than applied directly. Production live writes to GBrain are not yet supported. The bridge also provides fixture and test helpers so developers can work with the system without calling real external services, making it possible to develop and test locally using recorded responses.

The demos included in the repository run entirely without live network calls to GBrain, a language model, or any external deployment. They are designed to prove the integration mechanics using fixture data. The repository notes that GBrain versions not listed in its compatibility matrix are unverified, and it explicitly does not claim endorsement by either upstream project.

The README describes this as a proof-of-concept for a launch, and a STATUS.md file in the repository gives the current maturity and support state. Several planned features are listed as not yet implemented, including exposing usable tools via the ActiveGraph tools interface.