convez.ai
SkillBusca & PesquisaAvançado
yogsoth-ai

Hypothesis Formation

yogsoth-ai/hypothesis-formation

Transforma gaps em hipóteses testáveis e questões de pesquisa estruturadas

Ver no GitHub
InicialPluginScore · de 0 a 100

Motor de formação de hipóteses e questões de pesquisa orientado por objetivos. Transforma gaps e insights em hipóteses testáveis e questões de pesquisa precisas, com priorização multicritério, formulação estruturada e decomposição em sub-questões. Implementa três campanhas autônomas com orçamento quantitativo.

Instalação

$npx convez-ai yogsoth-ai/hypothesis-formation

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
3
Forks
0
Acompanhando
0
Saúde
50%
Último commit
há 5 meses
Versão
—
Issues abertas
0
Contribuidores
1

Casos de uso

  • Priorização multicritério de gaps de pesquisa usando AHP e otimização de portfólio
  • Derivação de hipóteses competitivas com critérios de falseabilidade e fronteiras
  • Formulação de questões de pesquisa calibradas (PICO/SPIDER/SPICE) com decomposição

Veredito

Skill robusto e bem-arquitetado para pesquisa acadêmica, com rigor metodológico (AHP, falseabilidade FINER). Exige compreensão de frameworks de pesquisa e integração em pipeline maior (DARE), não é para usuários casuais de pesquisa.

README

A tradução para o português deste README ainda está na fila.

A hypothesis is not a guess — it is a precisely engineered bridge between what we know and what we need to test. The difference between having an idea and having a hypothesis is the difference between pointing at the horizon and drawing a map to get there.

🧪 Hypothesis Formation

Goal-Driven Hypothesis & Research Question Formation Engine for Academic Research

🧭 Part of the De-Anthropocentric Research Engine. This repository is one of nine composable research packages that make up DARE — the full autonomous research-orchestration system. DARE bundles this package together with the others into a single self-contained clone, unified under one orchestrator. To use these skills as intended — with the spec-driven orchestrator and cross-package routing — clone the main repository rather than this repo alone.

Hypothesis Formation transforms upstream gaps and insights into testable hypotheses and precise research questions. You provide prioritized gaps — it ranks them, derives hypotheses with falsifiability criteria, and formulates framework-guided research questions with scope, success criteria, and sub-question decomposition. Three autonomous campaigns, each with quantitative budget enforcement.


⚡ What It Does

  • 🎯 Gap Prioritization — multi-criteria ranking of research gaps using AHP weighting, PiCMe assessment, pairwise comparison, portfolio optimization, and sensitivity testing
  • 🔬 Hypothesis Formulation — transforms insights into structured hypotheses via deductive, inductive, and abductive reasoning. Produces competing hypotheses with falsifiability criteria and boundary conditions
  • 📐 Research Question Formation — framework-guided formulation (PICO/SPIDER/SPICE/ECLIPSE), scope calibration, sub-question decomposition, and FINER criteria validation

🎯 Design Philosophy

🎖️ Campaign-Based Architecture

Each campaign is a self-contained research activity domain. The entry point routes to the correct campaign based on user intent, then the campaign autonomously selects strategies and executes with quantitative budget enforcement.

📐 Four-Level Hierarchy

ENTRY.md (root)
  → Campaign (3): self-contained research activity domain
    → Strategy: selected by research context and constraints
      → Tactic: multi-step orchestration pattern
        → SOP: single operation (import or subagent)

📏 Budget Enforcement

Every strategy embeds quantitative floors (dimensions evaluated, hypotheses produced, FINER criteria passed). Budget gates prevent premature exit. Context checkpoints after every strategy completion.

📖 兵法书 (Art of War) Pattern

Skills teach principles and provide default reference flows — CC reads, internalizes, and constructs its own approach for each specific research task. Hard constraints are minimal: Budget Gate, Minimum Yield, HARD-GATE, context-checkpoint.


🏗️ Architecture

┌───────────────────────────────────────────────────────────────┐
│  ENTRY POINT (ENTRY.md)                                       │
│  Routes to campaign based on research intent                  │
├───────────────────────────────────────────────────────────────┤
│  CAMPAIGN (3)                                                 │
│  gap-prioritization, hypothesis-formulation, research-question│
├───────────────────────────────────────────────────────────────┤
│  STRATEGY (15)                                                │
│  5 per campaign, selected by context                          │
├───────────────────────────────────────────────────────────────┤
│  TACTIC (10)                                                  │
│  3 + 4 + 3, orchestrate SOPs into workflows                  │
├───────────────────────────────────────────────────────────────┤
│  SOP (44)                                                     │
│  5 import + 2 shared + 37 campaign-specific                   │
│  Single-responsibility subagent operations                    │
└───────────────────────────────────────────────────────────────┘

📋 Campaign Overview

Campaign 1: Gap Prioritization

StrategyWhen to Use
multi-criteria-rankingDefault — 5-20 gaps, systematic multi-dimensional scoring
evidence-based-prioritizationGaps from rigorous literature review with strong evidence
stakeholder-weighted-rankingMultiple stakeholders with different priority weights
portfolio-optimization20+ gaps, portfolio-level risk/reward optimization
rapid-triage50+ gaps, quick filtering to manageable set

Campaign 2: Hypothesis Formulation

StrategyWhen to Use
deductive-hypothesis-generationMature theory exists, gap is theory-reality deviation
inductive-hypothesis-generationNo theory, but data patterns or empirical observations
abductive-hypothesis-generationAnomalous observations that defy existing explanations
competing-hypothesis-constructionNeed multiple alternatives to avoid confirmation bias
hypothesis-operationalizationHave directional hypothesis, need to formalize

Campaign 3: Research Question Formation

StrategyWhen to Use
framework-guided-formulationClear research type, standard framework available
scope-calibrationQuestion too broad or too narrow
decomposition-formulationComplex question, single study can't answer
comparative-formulationNeed to compare A vs B systematically
feasibility-constrained-formulationIdeal question exceeds available resources

🔧 Dependencies

DependencyWhat It Provides
web-browsingweb-search + web-research (2 import SOPs)
literature-enginepaper-overview + paper-search + paper-research (3 import SOPs)
subagent-spawningSubagent dispatch conventions
context-managementcontext-init + context-checkpoint protocol

MCP Servers Required

ServerPurpose
brave-searchWeb search for frameworks, methods, background
apifyDeep web scraping, Google Scholar
alphaxivPaper discovery, content, PDF Q&A
semantic-scholarPaper metadata, citations, recommendations

🚀 Quick Start

  1. Clone this repo alongside its dependencies:
git clone https://github.com/yogsoth-ai/hypothesis-formation.git
git clone https://github.com/yogsoth-ai/web-browsing.git
git clone https://github.com/yogsoth-ai/literature-engine.git
git clone https://github.com/yogsoth-ai/context-management.git
git clone https://github.com/yogsoth-ai/subagent-spawning.git
  1. Copy .mcp.example.json to .mcp.json and fill in your API keys

  2. Deploy skills to your Claude Code skills directory

  3. Use via the entry point:

/hypothesis-formation I have 10 gaps from my literature survey on [topic]. Help me prioritize and formulate hypotheses.

📊 Scale

ComponentCount
Campaigns3
Strategies15
Tactics10
Subagent SOPs37
Import SOPs5
Shared SOPs2
Total skill directories72

📄 License

Apache-2.0


A component of the De-Anthropocentric Research Engine, part of the Yogsoth AI ecosystem. Built by Pthahnix.

Ver o original no GitHub

Bom para

Relacionados