Search docs, scaffold code
@imqueue is an open-source TypeScript framework for building microservices that talk over a message queue instead of HTTP — you write a class, decorate the methods you want to publish, and callers get a fully-typed client generated from the running service. This server gives your assistant first-hand knowledge of it, so it writes real @imqueue code instead of plausible-looking code. search_docs covers the guides, the tutorial, the CLI manual, the articles, and every exported symbol of every @imqueue package that publishes an API reference — so you can ask a question in plain words ("how do I expose a service method?") or look up an exact signature ("RedisQueue.send", "IMQOptions.safeDelivery"). Results carry page URLs, and get_doc returns any page as plain markdown for reading and quoting. list_packages returns the current package catalogue with install commands, which is how an assistant picks the right one rather than guessing at a name. Where two packages cover the same ground — tracing, or the database layer — the results carry the rule for choosing exactly one, because installing both breaks quietly rather than loudly. scaffold_service generates an idiomatic service — an IMQService subclass with @expose()d, JSDoc-typed methods plus a bootstrap that starts it — from the method signatures you describe. A custom return type comes back with the @classType() and @property() decorators it needs, which matters because a field that lacks them reaches the generated client typed any and still compiles. scaffold_client shows how to generate and use the typed client for a service. Both return source text for you to review and paste; neither writes a file. All six tools are read-only. The server has no accounts, no authentication and no storage: each request is handled statelessly, and the only data it sees is the argument you send — a search query, a doc URL, a service name. Nothing is retained, and no conversation content is collected or used for training. Tools that would act on your own machine — driving the imq CLI, creating projects on disk, starting and stopping your services, reading their logs — are deliberately not part of this server, because a hosted server cannot reach your machine and should not claim to. They ship in the local install instead (npx -y @imqueue/mcp), and local_install_guide returns the setup steps for it.
Після підключення до @imqueue ChatGPT може передавати цьому застосунку релевантні чати й записи пам’яті для контексту ваших запитів. Використання цих даних застосунком @imqueue регулюється його умовами та політикою конфіденційності. Якщо у вас увімкнено Пам’ять, дані із застосунку можуть використовуватися, щоб завчасно надавати корисну інформацію чи рекомендації. ChatGPT завжди враховує ваші налаштування використання даних для навчання, зокрема щодо даних із підключених застосунків. Використання застосунків може супроводжуватися підвищеним ризиком. Керувати налаштуваннями або відключатися від застосунків можна будь-коли в налаштуваннях. Докладніше