Research Assistant

Research Insight and Exploration Assistant — an AI co-pilot for discovering, summarizing, and drafting research content.

Getting Started

Install dependencies once you have network access:

npm install

Run the local development server (Vite with hot module reloading):

npm run dev

If you have deno installed, you can also use deno:

deno install
deno run dev

Build and preview a production bundle:

npm run build
npm run preview

Execute automated checks:

npm run lint
npm run test
npm run test:coverage

Project Layout

• src/main.ts bootstraps Vue, PrimeVue, Tailwind, and Pinia with persisted state.
• src/layout/AppShell.vue holds the fixed 3rem header, adjustable workspace, and 1rem footer.
• src/features/reading implements PDF ingestion, pdf.js rendering, metadata summaries, and the co-pilot chat.
• src/features/writing provides the markdown library, CodeMirror editor, contextual instruction dialog, and advisor chat.
• src/stores contains Pinia stores for layout, reading artifacts, writing drafts, and shared chat history.
• src/lib/agents/client.ts wraps @openai/agents, falling back to simulated responses until an API key is provided.

Environment Configuration

Create .env.local from .env.example and supply your OpenAI key:

VITE_OPENAI_API_KEY=sk-...

All user data (uploads, summaries, chats) is persisted to localStorage; clear browser storage to reset the workspace.

Feature Highlights

• Split workspace with draggable columns, vertical tab selectors, and PrimeVue menubars per panel.
• Reading assistant supports multi-PDF uploads, pdf.js rendering, editable metadata, and iterative summaries.
• Writing advisor manages multiple markdown drafts, offers CodeMirror editing, and exposes right-click instructions that inject AI-generated prose.
• Co-pilot chat threads are scoped by document and survive reloads through Pinia persistence.