Structure

Layers are inferred from where files sit in the tree, not from reading the code. A project that names its directories unconventionally will read oddly here — open the file browser to check anything the diagram implies.

Languages

Share of indexed source by file size. Binary and vendored files are excluded.

Dependencies

Declared in the repository’s manifests at the indexed commit. A declared package is not proof it is used, and runtime dependencies are listed first.

Feature verification

• Agora real-time video/voice callingVerified

  Agora's real-time voice and video capabilities power the video call platform

  Claimed on readmehigh confidence

  • frontend/src/components/Basics.jsx:1 — Uses agora-rtc-react hooks (useJoin, useLocalCameraTrack, useLocalMicrophoneTrack, usePublish, useRemoteUsers) and renders LocalUser/RemoteUser video tiles, a functioning join-channel UI
  • frontend/package.json:16 — agora-rtc-react and agora-rtc-sdk-ng are real dependencies
  • frontend/src/routes/Dashboard.jsx:46 — Basics (the Agora call component) is wired into the Dashboard page
• FastAPI backend for API endpointsVerified

  The backend is implemented using Python FastAPI to handle our API endpoints

  Claimed on readmehigh confidence

  • backend/server.py:33 — FastAPI app is created with CORS middleware and endpoints /query-agent and /create-user
• Fetch.AI agents for notes generation and vector searchCode-supported

  Fetch.AI agents simplify the process of generating notes and vector search

  Claimed on readmemedium confidence

  • backend/meeting_agents.py:27 — Defines notes_agent and vector_search_agent using uagents.Agent with on_query handlers for notes generation and vector search, and a Bureau to run them
  • backend/vectorDB.py:79 — use_vector_search_agent queries vector_search_agent, showing intended agent orchestration, but this function is only invoked from create_and_insert_embeddings, which is not called by any API route or frontend code
• Gemini AI note/summary generationCode-supported

  Google's Gemini AI generates a concise summary and detailed notes (decisions, action items, deadlines, next steps)

  Claimed on readmemedium confidence

  • backend/meeting_agents.py:108 — Configures genai and calls gemini-1.5-pro-latest with a prompt requesting summary, decisions, action items, deadlines, and next steps, matching the claim text closely
  • backend/meeting_agents.py:80 — handle_notes is only triggered by an on_query handler on notes_agent; no code in server.py or the frontend ever calls this agent, so it is not reachable end-to-end
• OpenAI text embeddings generationCode-supported

  We also utilized the OpenAI API to generate text embeddings on our vector database

  Claimed on readmemedium confidence

  • backend/vectorDB.py:29 — Calls OpenAI client.embeddings.create with model text-embedding-ada-002 to generate embeddings and stores them, but this code path is never invoked from any reachable API endpoint (only from __main__ or unused helper functions)
• SingleStore vector database for retrievalCode-supported

  SingleStoreDB vector database allows fast and accurate retrieval of information from large datasets

  Claimed on readmemedium confidence

  • backend/vectorDB.py:39 — Inserts OpenAI text embeddings into a SingleStore myvectortable via singlestoredb connection
  • backend/meeting_agents.py:60 — vector_search agent runs a SingleStore dot-product similarity query (v <*> @qv) with an HNSW_PQ vector index, implementing vector search
  • backend/server.py:49 — The only exposed API endpoint that could trigger this path, /query-agent, calls an undefined function agent_query (never imported or defined anywhere in the file), so the vector search path is not reachable through the running server
• TypeScript/TailwindCSS frontend with clean minimal UICode-supported

  The front end of MeetAI is built using TypeScript and Tailwind.css to create a clean and minimalistic user interface

  Claimed on readmemedium confidence

  • frontend/tailwind.config.js — TailwindCSS is configured and used throughout components (e.g. Basics.jsx uses Tailwind utility classes)
  • frontend/tsconfig.json — TypeScript config exists, but most source files are .jsx, not .tsx (only frontend/src/components/ui/navigation-menu.tsx and frontend/src/lib/utils.ts use TypeScript), so the app is only partially TypeScript
• User account creation / sign-up-sign-inCode-supported

  Implied by Sign-in/Sign-up routes and Clerk dependency for a platform with hosts

  Claimed on Devpostlow confidence

  • frontend/src/routes/Sign-in.jsx — Dedicated sign-in route file exists
  • frontend/src/routes/Sign-up.jsx — Dedicated sign-up route file exists, and @clerk/clerk-react is a listed dependency, but full auth wiring was not traced end-to-end
  • backend/server.py:57 — A separate /create-user endpoint inserts directly into a SingleStore users table, unrelated to Clerk, suggesting inconsistent/duplicated user creation logic
• Real-time feedback and notes during the call (host decision assistance)Claimed only

  MeetAI generates a concise summary and detailed notes all while the video call is going on, enabling a feedback-driven cycle for the host

  Claimed on readmehigh confidence
• View Reports / notes viewing UIClaimed only

  Implied by 'detailed list of notes' surfaced to the host during the meeting

  Claimed on Devposthigh confidence

  • frontend/src/routes/Dashboard.jsx:48 — The 'View Reports' tab only renders a static placeholder heading 'Here are the reports.' with no data fetching or connection to backend notes

An AI agent derived these features from the project’s Devpost page and readme, then searched the code for each one. Verified features are backed by cited code; claimed-only features had no supporting code, which is not by itself proof a feature is missing.