This project was not completed, and is not being developed.
For the memory-bank format — the specification and a starting skeleton — see Supervertaler/Supervertaler-SuperMemory.
A standalone cross-platform desktop app (Python + PyQt6) that would sit beside any CAT tool and read the same memory bank as Supervertaler for Trados — chat over your accumulated client, terminology, domain and style knowledge, process an inbox of raw material into structured articles, run a health check, and distil TMX/DOCX/PDF/TBX files into draft notes.
It reached a working prototype and stopped there.
The problem it existed to solve — using a memory bank outside Trados Studio — was solved a different way.
Since v18.20.146 Supervertaler for Trados exposes memory banks over its MCP server, so any MCP client can read your bank directly: Claude Desktop, Claude Code, or anything else that speaks the protocol. You get chat over your knowledge without a second application to install, update and maintain, and using whatever AI subscription you already pay for.
Against that, a standalone app would have needed its own chat interface, its own provider plumbing, its own release process — to arrive at roughly the same place. So it stays here as reference rather than being finished.
The code remains readable and MIT-licensed, should anyone want it:
supervertaler_assistant/memory_bank.py— memory-bank reader: frontmatter parsing, relevance scoring, context buildingsupervertaler_assistant/agents/— the compile, lint, distil and query agentssupervertaler_assistant/llm.py— a thin LiteLLM wrapper covering ~100 providers
The format specification and skeleton that used to live here have moved to Supervertaler-SuperMemory, which is where they are maintained.
MIT — see LICENSE.