This project provides a portable agent skill, "good-question," designed to help researchers systematically refine and sharpen their research questions. It targets individuals with nascent research ideas, literature gaps, or proposal sketches, offering a structured approach to transform vague interests into robust, defensible questions. The primary benefit is enabling users to assess a question's importance, feasibility, and potential impact, thereby improving research direction and decision-making.

How It Works

The skill operates by first diagnosing the user's current research stage (e.g., broad interest, literature gap, proposal). It then employs a rigorous, multi-step workflow that includes inferring an appropriate working mode (mentor, reviewer, collaborator, or grant advisor), building concise domain-specific briefs when necessary, and leveraging "first principles" thinking as a calibration layer. This calibration checks the question's stakes, underlying assumptions, rival hypotheses, falsifiability, and evidence boundaries, without bypassing essential source audits or field-specific evidence norms. Candidate questions are generated using structured lenses and converged upon using criteria such as importance, feasibility, and falsifiability, culminating in a simulated "reviewer-style" stress test to identify potential objections. This methodical process distinguishes it from simple brainstorming tools by ensuring questions are not only novel but also significant, testable, and grounded.

Quick Start & Requirements

• Installation: Clone the repository into your agent's skill directory:
  • Codex: git clone https://github.com/Rimagination/good-question.git ~/.codex/skills/good-question
  • Claude: git clone https://github.com/Rimagination/good-question.git ~/.claude/skills/good-question
  • Other agents can utilize SKILL.md as the primary workflow and references/ for method cards.
• Prerequisites: Requires an agent environment (e.g., Codex, Claude). No other specific software or hardware dependencies are listed.
• Links: Field Playbooks are available at docs/field-playbooks.md.

Highlighted Details

• Outputs a structured "Good Question Card" detailing the working title, core research question, significance, challenged assumptions, competing hypotheses, proposed pilot experiments, falsification conditions, and potential reviewer objections.
• Integrates established research methodologies and principles from figures like Alon, Platt, Alvesson & Sandberg, and Hamming.
• Provides field-specific playbooks tailored for domains including ecology, remote sensing, AI4Science, social science, biomedicine, humanities, and engineering systems.
• Employs "first principles" as a calibration tool to expose fundamental constraints and assumptions, enhancing the rigor of question formulation.

Maintenance & Community

No specific details regarding maintainers, community channels (e.g., Discord, Slack), or project roadmap are provided in the README. The project is hosted on GitHub at Rimagination/good-question.

Licensing & Compatibility

The provided README does not specify a software license. This omission requires further investigation to determine usage rights, particularly for commercial applications or integration into closed-source projects.

Limitations & Caveats

The skill is not an exhaustive knowledge base and does not fabricate field consensus. It explicitly labels inferred information versus source-backed evidence and requires enhanced retrieval for insufficient context. It will critically evaluate ideas, potentially suggesting rewrites, shelving, or discarding those lacking novelty, audience, falsifiability, or learning potential from negative outcomes. Its effectiveness may be limited in highly specialized domains without additional user-provided context.