How rastogivaibhav/moolbase works
MoolBase is a persistent evidence-aware memory and reasoning substrate for agents. Its principal workflow is to ingest observations into durable graph memory, retrieve evidence, assess competing explanations and contradictions, then expose a governed answer and an inspectable receipt. The graph also preserves documented policy-learning and vector-retrieval capabilities. Several implementation connections are based on the README rather than sampled call sites; those are marked dashed or omitted where the wiring is uncertain.
Main entry points: hypokosh_runtime.cpp Governed reasoning · db.cpp Persistent graph DB · learning.cpp Learning episodes · graphenedb_client.py Python client
Components 19
Application interfaces
- Python client
clients/python/graphenedb_client.py - HTTP server
tools/graphenedb_server.cpp - Native API
src/c_api.cpp
Evidence memory
- Persistent graph DB
src/db.cpp - Vector indexes
include/graphene/kdtree_index.hpp - Relations and ontology
src/relation_ontology.cpp - Entity resolution
src/entity_resolution.cpp
Reasoning and governance
- Governed reasoning
src/hypokosh_runtime.cpp - Evidence expansion
src/dialectic.cpp - Evidence paths
src/fiber_bundle.cpp - Stability assessment
src/stability_critic.cpp - Epistemic control
src/epistemic_control.cpp - Path verification
src/path_verifier.cpp - Model world
src/model_world.cpp - Decision receipts
src/epistemic_receipt.cpp
Policy learning
- Learning episodes
src/learning.cpp - Policy evaluation
src/learning.cpp
Other components
- Agent application
- Governed answer
Connections 21
- Agent application → Python client: submits requests
- Python client → HTTP server: calls HTTP
- Agent application → Native API: drives native API
- HTTP server → Persistent graph DB: serves memory
- Native API → Persistent graph DB: accesses memory
- Persistent graph DB → Vector indexes: uses indexes
src/db.cpp - Persistent graph DB → Relations and ontology: stores graph relations
- Persistent graph DB → Entity resolution: supports resolution
- Agent application → Governed reasoning: requests reasoning
- Governed reasoning → Persistent graph DB: reads evidence
- Governed reasoning → Evidence expansion: expands evidence
- Evidence expansion → Evidence paths: builds paths
- Governed reasoning → Stability assessment: assesses stability
- Governed reasoning → Epistemic control: governs decisions
- Governed reasoning → Path verification: verifies paths
- Governed reasoning → Model world: updates world
- Governed reasoning → Decision receipts: records decision
- Governed reasoning → Governed answer: returns status
- Agent application → Learning episodes: records outcomes
- Learning episodes → Persistent graph DB: persists episodes
- Policy evaluation → Learning episodes: evaluates episodes
src/learning.cpp