Guide hypothesis-driven research from literature survey to publication with locked protocols.
npx skills add https://github.com/AmberLJC/meta-research --skill meta-research-amberljc
Please help me install this Agent Skill. Skill: meta-research Source: https://github.com/AmberLJC/meta-research/tree/main Command: npx skills add https://github.com/AmberLJC/meta-research --skill meta-research-amberljc
Research projects in AI and science often fail not from lack of effort, but from lack of rigor—skipping literature review, generating untestable hypotheses, running experiments without locked protocols, and failing to track decisions. This skill enforces a structured, hypothesis-driven workflow that prevents common pitfalls like bias, irreproducibility, and researcher degrees of freedom.
Invoke the meta-research skill with your research question to begin a structured literature survey and hypothesis generation workflow.
High-intent search queries and answers about installing and using this skill.
Maintain a structured hypothesis tree in YAML format to track decision branches. This enforces rigorous reproducibility by linking chronological research logs directly to locked experiment protocols and pre-committed analysis plans.
A hypothesis-driven research workflow is a 6-phase loop from literature survey to reflection. You need it when conducting scientific or AI experiments requiring structured hypothesis testing, bias mitigation, and strict reproducibility.
Design locked experiment protocols by pre-committing analysis plans before execution. This prevents researcher degrees of freedom by enforcing clear confirmatory and exploratory boundaries within your structured experiment design.
Yes, you can use this workflow for small model fine-tuning. It guides you to survey SOTA methods, generate testable hypotheses, design ablation plans, execute sanity checks, and reflect on whether to pivot or conclude.
Yes, the research workflow supports autonomous execution via Clawbot Executor mode and periodic strategic review via Research Advisor mode. This dual operation enables both task execution and continuous rigor auditing.
AI research projects fail by skipping literature review and running experiments without locked protocols. Prevent this by enforcing a structured workflow that mitigates bias, tracks decisions chronologically, and requires pre-registered analysis plans.
WanLanglinGenerate three to five testable hypotheses with predictions and experimental plans.
randDiscover ML and AI skills during model training and inference workflows.
soheunyiPlan reproducible experiments with explicit controls, seeds, and evaluation protocols.
Unified-Field-Theory-ResearchExecute quick research tasks in GPD projects with atomic commits.
