dsh-arch-advisor

When you discuss system design with an AI, the model typically relies on training-time memory to produce a generic, one-paragraph suggestion. It lacks real-world case backing and a concrete document skeleton you can actually ship. dsh-arch-advisor closes that gap: it wraps the open-source awesome-architecture knowledge base (40 tutorials, 31 real-system templates, 6 end-to-end cases) into seven callable arch_* tools so the model moves from vague impressions to citing specific templates, cases, and methodologies step by step.

The seven tools fall into three groups. Read: arch_roadmap gives a learning-path overview, arch_search indexes the catalogue by Chinese or English keywords, and arch_read opens any article with a table of contents, preview, and full-text access in either language. Discuss: arch_ask maps your one-line idea to the most relevant tutorials, templates, and cases and returns an actionable route; arch_design walks you through six stages—clarify requirements, pin constraints and quality attributes, choose structure, document key decisions as ADRs, sketch an evolution roadmap, and run a production-grade review—with each stage discussed line by line. Write: arch_docs supplies standard section skeletons plus completion checklists for PRD, HLD, LLD, DBD, and architecture documents; arch_version enforces a current-version plus archives/ directory convention so every finalized revision leaves a permanent, traceable snapshot.

It is built for people actively designing systems, developers and product managers who need to produce requirement or design documents, engineers preparing for system-design interviews, and anyone who wants an AI to deliver grounded, methodical, and shippable architecture work products.