CatalogMesh

CatalogMesh is an AI workspace for product catalog operations, covering grouping, review, SKU matching, exports, storage, and guarded automation. It turns a raw chronological product shoot into reviewed product groups, catalog/SKU matches, exports, cloud-storage copies, and controlled connector actions, while the original photos stay untouched. Human review remains the authority for catalog identity and externally visible publication.

The product provides a sorting engine with multi-angle product grouping, crash-safe SQLite resume, progress/ETA, and reports. It supports cloud AI through Gemini, OpenAI, and Anthropic provider pools, local AI through Ollama, and local evidence such as embeddings shadow mode, OCR, barcode evidence, calibration, and a Hybrid Routing Lab. A Review Center offers non-destructive merge/split/move, corrections, approval state, and audit history. SKU matching produces ranked deterministic catalog candidates with mandatory human confirmation. Exports include safe offline Shopify draft and neutral PIM export profiles. Storage Center uses rclone for local-first copy, dry-run, and manual mirror to configured cloud remotes, with safe automatic post-run copy. Automation Center is generated from the same automation CLI parser, so command parity is tested automatically. Connectors provide approval-aware Shopify, Akeneo, and Odoo execution with connector-specific safety boundaries.

CatalogMesh requires Python 3.10+ and can be installed via pip. The normal sorter GUI and CLI share the same processing engine. Remote mutation commands require approval and single-use reservation artifacts plus an additional GUI confirmation phrase. The current stable release is v3.3.2.