claude-scholar: a semi-automated research assistant for academic research and software development

What it solves

Claude Scholar is a semi-automated research assistant designed to accelerate the heavy, repetitive, and structure-sensitive parts of academic research and software development. It specifically targets the friction in literature organization, note-taking, experiment analysis, reporting, and writing, while ensuring that the human researcher remains the primary decision-maker for high-level judgments and hypotheses.

How it works

It functions as a set of configurations, skills, agents, and hooks that integrate with AI CLI tools like Claude Code, Codex CLI, Kimi Code CLI, and OpenCode. The system routes research through a traceable pipeline: question -> evidence -> experiment -> analysis -> claim -> writing. It leverages integrations with Zotero (for literature management via MCP) and Obsidian (for project knowledge management) to maintain a structured, durable record of research progress.

Who it’s for

• Computer science and AI/ML researchers who frequently switch between coding, reading papers, and writing.
• Research engineers and graduate students seeking a structured workflow for implementation and analysis.
• Software-heavy academic projects that require reproducible project memory and CLI automation.

Highlights

• Multi-Platform Support: Compatible with Claude Code, Codex CLI, Kimi Code CLI, and OpenCode.
• End-to-End Pipeline: Specialized skills for ideation, literature synthesis, results analysis, and publication-grade writing.
• Tool Integrations: Deep integration with Zotero for paper import/management and Obsidian for vault-based knowledge management.
• Scientific Production: Includes pubfig and pubtab for creating publication-ready figures and tables.
• Evidence-Gated Workflow: Uses a research-contract.md to track claim strength and evidence records before promoting findings to final writing.