Implement step 17 modules: threads/messages, notifications with automation (submission.created, idea.created, grant_call.opened), innovation test feedback, admin panel (submissions, verify, metrics, grant call status), Middleman adaptation and partnerships. Split multi-type files into one type per file and replace object-valued dictionaries with string-valued maps in data contracts (migrations 0006/0007).
Implement the idea creator and application generator: idea status workflow, social canvas card, grant call windows with 409 when closed, and an LLM-backed idea assistant with offline fallback.
Add a pluggable embedding provider with an offline deterministic default, vector indexing for problems and innovations, cosine semantic ranking, similar cases and solution suggestions, area filtering and an admin reindex endpoint. Refresh the index when innovations or submissions change.
Add knowledge item and innovation CRUD with search/filtering, published-only visibility for non-staff, admin publish/update, challenge dictionary, problem aggregation with trends (admin only) and canonical embedding text preparation for matchmaking.
Add submission CRUD with pagination and role scoping, challenge/location binding, automatic keyword extraction, a validated status workflow, and domain events for matchmaking and notifications. Create a problem record together with each submission.
Add signup/signin with backend-issued JWT (Argon2 passwords in SurrealDB), role-based authorization policies, per-user data access levels, consent register and read audit log. Protect user endpoints and record sensitive reads.
Split the backend into API host, Domain, Application and Infrastructure projects. Add Swagger/OpenAPI per API version, Asp.Versioning (query/header), ProblemDetails with a global exception handler and SurrealQL migrations applied at startup (namespace/database ensured). Infrastructure owns the SurrealDB client, settings and DI.
Rafał Tyński