Finverity Chat

Codebase landing

loadingMCP GuideOpen Chat Workspace

Internal Retrieval Surface

MCP-ready

One control room for backend, frontend, and infra code.

Retrieval-first chat with visible evidence, repo-aware scope control, and a model layer that can later be shared with other LLM clients through MCP.

Launch ChatStart With SSO Route Trace
Which backend routes are responsible for SSO settings initiated from the frontend?Trace the flow from frontend action to backend service for tenant authentication changes.Which infrastructure assets define runtime dependencies for workflow-service?Show me where gateway auth changes propagate into frontend apps.
Vector Storeqdrant

Qdrant-backed retrieval surface with repo-level filters.

Embeddingsdeterministic

Current retrieval driver for chunk similarity and reranking.

LLMopenai

Retrieval-only fallback mode.

ScopeAll repositories

Backend Workspace · Frontend Workspace · Infrastructure Workspace

Repository Coverage

Connected Workspaces

The landing page is now separate from the chat workspace.

Backend Workspacebackend

NestJS and microservice backend repositories.

/Users/mpx/WebstormProjects/finverity_backend_workspace_1
Frontend Workspacefrontend

Angular platform app and shared authentication libraries.

/Users/mpx/WebstormProjects/finverity_frontend_workspace
Infrastructure Workspaceinfrastructure

Infrastructure code, runtime definitions, and deployment assets.

/Users/mpx/CursorProjects/finverity_infrastructure_workspace

Working Model

How The Chat Operates

Semantic retrieval, graph-aware enrichment, and grounded answer generation stay behind the chat page.

Step 1Choose intent and scope

Route the question to semantic retrieval, graph tools, or both.

Step 2Retrieve grounded evidence

Collect file paths, routes, symbols, and graph insights before generation.

Step 3Answer inside the workspace

The `/chat` page keeps repository scope and retrieved context available but now hideable.