Python Embedding Sidecar Setup¶
The Neuralgentics TUI can use a Python gRPC sidecar for BGE-Large (1024-dim) embeddings. Without the sidecar, memory operations still work — they use noop (384-dim) embeddings instead. The sidecar is an advanced feature for users who need high-quality semantic search.
Quick Start (Source Tree)¶
If you cloned the repo:
This creates a venv, installs dependencies, and starts the sidecar on /tmp/neuralgentics-embed.sock. The TUI detects it automatically.
Manual Setup¶
- Create a venv in the sidecar directory:
cd packages/memory/cmd/embedding-sidecar
python3 -m venv .venv
.venv/bin/pip install -r requirements.txt
- Start the sidecar:
- Point the TUI to it (if not using the default socket):
Installed Binary (No Source Tree)¶
Set NEURALGENTICS_SIDECAR_DIR to tell the TUI where to find the sidecar:
Or install the sidecar to the standard location:
mkdir -p ~/.neuralgentics/share/neuralgentics/sidecar
# Copy the sidecar source there, then create venv + pip install
The TUI checks $NEURALGENTICS_SIDECAR_DIR, then $NEURALGENTICS_INSTALL_PREFIX/share/neuralgentics/sidecar/, then the source-tree path.
GPU Acceleration¶
Set NEURALGENTICS_EMBED_DEVICE=cuda before starting the sidecar:
Stopping¶
Or kill the PID in /tmp/neuralgentics-embed.pid.