Scibudy Documentation

Scibudy is a Codex-native research assistant for scholarly search, reusable paper libraries, full-text analysis, local evidence retrieval, and journal-style diagnostics.

Core surfaces

• CLI: scibudy
• MCP server: scibudy-mcp
• Installer: scibudy-install
• UI: scibudy ui --open

Five-minute path

npx scibudy-install --profile base
scibudy doctor --json
scibudy install-codex
codex mcp get research
scibudy search "simulation-based calibration"
scibudy ui --open

Documentation map

• Prerequisites
• Installation
• Quickstart
• Codex MCP setup
• CLI and MCP
• Examples
• Configuration
• GPU local models
• Library workflow
• Journal style analysis
• Troubleshooting
• Development
• Releasing

Release links

• GitHub Releases: https://github.com/ONEMULE/scibudy/releases
• npm installer: https://www.npmjs.com/package/scibudy-installer
• PyPI package: https://pypi.org/project/scibudy/

Architecture summary

Scibudy is Python-first:

• research_mcp/ contains the runtime, MCP server, CLI, and analysis stack
• web/ contains the UI source and distributable build artifacts
• bin/scibudy-install.mjs is the npm/bootstrap wrapper
• user runtime state is stored under the app home, not in the source tree

Product model

• base profile: CPU-safe runtime and Codex integration
• analysis profile: full-text analysis workflow
• gpu-local profile: local model environment
• full profile: all layers together