C++ SCHX ยท rev 4
Start here ยท v0.24.0

ABRIA overview

What ABRIA 0.24.0 is, how ABR differs from a foundation model, and how the native C++ and Python layers fit together.

SCHX / FEATURE LAYER

Source map

  • python/abria_server/__init__.py
  • python/abria_server/api.py
  • python/abria_server/abr_runtime.py
  • cpp/include/abria/abr.hpp
SCHX / FEATURE LAYER

ABRIA 0.24.0 in one view

The uploaded runtime identifies itself as ABRIA 0.24.0. The Python service exposes the public API and orchestrates native C++ cognition, per-user ABR fabric, memory, tooling, workspaces, research and optional learning layers.

In ABR mode, the primary runtime does not require a pretrained foundation checkpoint. The native AbrBrain grows a user-specific cognitive fabric from experiences: concepts, relations, constructions and reusable relation-composition programs.

  • Python runtime version: 0.24.0
  • Native version is reported separately by /api/health
  • Per-user ABR state is isolated by user identity
  • Guest sessions are isolated but non-persistent
SCHX / FEATURE LAYER

Two intelligence paths

ABRIA contains more than one execution path. ABR mode is the personal cognitive-fabric runtime; the broader codebase still contains optional neural/foundation components used by other modes and experiments.

The docs call this distinction out explicitly so 'ABRIA has neural code' is not confused with 'ABR itself is a neural network'.

  • ABR: concepts, relations, constructions, programs, theories/abstractions and personal fabric
  • Optional neural_v2/foundation/online-learning modules remain separate layers
SCHX / FEATURE LAYER

What the docs cover

These pages document the architecture from the uploaded C++ and Python trees, plus the modern ABRIA workspace UI pattern already used by the project.

  • C++ native brain and cognitive fabric
  • Python API, identity and AETHER bridge
  • Workspace/file/code intelligence
  • PABR personal twin and media
  • Safety, research, tools and verification