Crate ospipe 

OSpipe

RuVector-enhanced personal AI memory system integrating with Screenpipe.

OSpipe captures screen content, audio transcriptions, and UI events, processes them through a safety-aware ingestion pipeline, and stores them as searchable vector embeddings for personal AI memory recall.

Architecture

Screenpipe -> Capture -> Safety Gate -> Dedup -> Embed -> VectorStore
                                                            |
                                     Search Router <--------+
                                     (Semantic / Keyword / Hybrid)

Modules

• capture - Captured frame data structures (OCR, transcription, UI events)
• storage - HNSW-backed vector storage and embedding engine
• search - Query routing and hybrid search (semantic + keyword)
• pipeline - Ingestion pipeline with deduplication
• safety - PII detection and content redaction
• config - Configuration for all subsystems
• error - Unified error types

Modules

capture

Capture module for processing screen, audio, and UI event data.

config

Configuration types for all OSpipe subsystems.

error

Unified error types for OSpipe.

graph

Knowledge graph integration for OSpipe.

learning

Continual learning for search improvement.

persistence

JSON-file persistence layer for OSpipe data.

pipeline

Ingestion pipeline with deduplication.

quantum

Quantum-inspired search acceleration.

safety

Safety gate for content filtering and PII redaction.

search

Query routing and hybrid search.

server

Lightweight HTTP REST API server for OSpipe.

storage

Vector storage, embedding engine, and trait abstractions.

wasm

WASM bindings for OSpipe.