research-primitives

Catalog research primitive operations for scholarly evidence workflows in the MCP toolkit.

Updated Aug 27, 2026
One-click install
npx skills add https://github.com/Biajin-PKU/research-harness --skill research-primitives
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: research-primitives
Source: https://github.com/Biajin-PKU/research-harness/tree/main/codex-skills/research-primitives
Command: npx skills add https://github.com/Biajin-PKU/research-harness --skill research-primitives

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This guide helps researchers and engineers quickly understand and leverage the core MCP research primitive operations without hunting through scattered documentation.

Core Features & Use Cases

  • Comprehensive reference to all 9 primitives: paper_search, paper_ingest, paper_summarize, claim_extract, evidence_link, gap_detect, baseline_identify, section_draft, consistency_check.
  • Enables end-to-end workflows for retrieval, extraction, generation, and verification of scholarly evidence across papers and topics.
  • Use cases include literature reviews, evidence assembly with provenance, and drafting sections anchored to linked claims.

Quick Start

Read this guide and try a small workflow that uses paper_search, paper_ingest, paper_summarize, and claim_extract on a sample topic.

Frequently Asked Questions about research-primitives

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
What are research primitives in the MCP toolkit?

Research primitives are core operations for retrieving, extracting, generating, and verifying scholarly evidence across papers and topics. They provide discoverability of inputs, outputs, and MCP command integration to support end-to-end literature workflows.

How do I extract claims and link evidence from multiple papers?

To extract claims and link evidence, use the claim_extract and evidence_link primitives. These operations support evidence assembly with provenance, enabling you to link specific scholarly claims directly to their source literature across multiple papers.

Can I draft research sections anchored to linked claims using MCP?

Yes, you can draft sections anchored to linked claims using the section_draft primitive. This operation generates text for research sections while maintaining provenance by anchoring the drafted content directly to previously extracted and linked evidence.

What is the best way to start a literature review workflow with MCP?

To start a literature review workflow, sequentially apply the paper_search, paper_ingest, paper_summarize, and claim_extract primitives. This workflow allows you to retrieve, ingest, summarize, and extract claims from papers on a sample topic.

Does the MCP toolkit support identifying research gaps and baselines?

Yes, the MCP toolkit supports identifying research gaps and baselines using the gap_detect and baseline_identify primitives. These operations help researchers discover missing evidence within literature and establish baseline references for scholarly topics.