skills.rest
yslee5005
yslee5005COMMUNITY
Explore all skills by this author →
GitHub

research-protocol

Standardize evidence sourcing, fact-inference separation, and critique in research workflows.

Updated Apr 5, 2026
Search Tags:#research#verification#evidence#sourcing#not-found#fact-vs-inference#hallucination-defense
One-click install
npx skills add https://github.com/yslee5005/app-library --skill research-protocol-yslee5005
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: research-protocol
Source: https://github.com/yslee5005/app-library/tree/main/apps-internal/agent-hub/.claude/skills/research-protocol
Command: npx skills add https://github.com/yslee5005/app-library --skill research-protocol-yslee5005

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Provides a standardized research protocol for the Agent Hub team to gather, cross-examine, and critique evidence, defining sourcing standards, separating facts from inferences, and defending against hallucination.

Core Features & Use Cases

  • Multiple independent sources: Prefer primary or official sources over secondary references to support conclusions.
  • Source tagging for claims: Tag every non-obvious claim with its source to distinguish evidence from inference.
  • Fact-vs-inference separation and cross-checking: Explicitly separate what is known from what is inferred and challenge conclusions to falsify them.
  • Recency tracking and not-found labeling: Note the date of sources and clearly indicate when information is not found.

Quick Start

Provide a structured, source-backed evaluation by gathering multiple independent sources, tagging every non-obvious claim with its source, separating fact from inference, cross-checking and falsifying findings, noting recency, and clearly marking not-found results.

Frequently Asked Questions about research-protocol

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

FAQPage Schema
How do I prevent AI hallucination during research and evidence gathering?▼

To prevent AI hallucination during research, this protocol requires explicit sourcing for non-obvious claims, fact-vs-inference separation, and clear labeling of not-found results to ensure trustworthy outputs.

What is the best way to separate fact from inference in a research report?▼

Separating fact from inference requires tagging every non-obvious claim with its source, explicitly distinguishing known evidence from inferred conclusions, and cross-checking findings to falsify hypotheses.

How do I structure a research protocol to cross-examine multiple independent sources?▼

Structure a research protocol by preferring primary or official sources over secondary references, tagging claims with their sources, cross-examining evidence, tracking recency, and surfacing not-found results.

How do I track source recency and handle not-found results in evidence reporting?▼

Track source recency by noting the date of sources used, and handle not-found results by clearly indicating when specific information is not found to maintain transparency in evidence reporting.

Does this research verification protocol work for analysts gathering evidence without coding dependencies?▼

Yes, this research verification protocol works for analysts and critics without coding dependencies, providing a standardized framework for sourcing, evidence handling, and critique to ensure trustworthy outputs.

Related Skills

Explore More Education & Research Skills→
yu13130122297yu13130122297

scout

Identify uncertainties in research framing, datasets, metrics, and baselines.

Community
Intermediate
ylxmf2005ylxmf2005
11

parallel-researcher

Aggregate and verify evidence from multiple sources into lane-level reports.

Community
Advanced
Bhuvansai-16Bhuvansai-16
2

deep-reasoning-agent

Verify research drafts through citation validation, fact-checking, and content quality assessment.

Community
Advanced
yudameyudame
2

perplexity-deep-research

Automate Perplexity Deep Research API calls for academic literature reviews.

Official
Intermediate