Skip to content

UMD-ARLIS/open-analyst

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

217 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Open Analyst

Headless, project-oriented analyst workspace.

Open Analyst runs as:

  • Browser UI (Vite/React)
  • Headless API service (Node HTTP)

No Electron runtime is required.

Quick Start

Requirements:

  • Node.js 20+
  • npm

Install:

npm install

Run full stack (UI + headless API):

npm run dev:stack

Run services separately:

npm run serve:headless   # API on :8787
npm run dev:web          # UI on :5173

Build and Test

npm test -- --run
npm run build

Core Workflow

  1. Create/select a project in the left sidebar.
  2. Create/select a collection for that project.
  3. Start or continue a task in chat.
  4. Use fetch/search/research tools during chat.
  5. Captured sources are stored in the active project collection and available for RAG.
  6. Reopen old tasks from project task history.

Key Capabilities

  • Project-first workspace (not task-only)
  • Collections and source/document management
  • Agentic tool execution during chat
  • Deep research tool for multi-step source discovery and synthesis
  • Deep search + retrieval over project documents
  • Source capture and indexing from web/PDF/research APIs
  • MCP server configuration
  • Skills management (builtin + custom install)
  • Credentials and logs management

Important Paths

  • UI: src/renderer
  • Headless API: scripts/headless-server.js
  • Project/data store: scripts/headless/project-store.js
  • Architecture + status: ARCHITECTURE_MAP.md

Documentation Scope

Documentation is intentionally consolidated into two files:

  • README.md
  • ARCHITECTURE_MAP.md

About

Forked Open Source Implementation of a Claude Cowork-like Analyst AI Assistant

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages

  • Python 40.2%
  • TypeScript 30.3%
  • JavaScript 24.9%
  • HTML 4.0%
  • Other 0.6%