K-Dense-AIOfficial·15 Agent Skills Included

science-superpowers

Rigorous research question framing, pre-registration, and reproducible analysis

Guides research agents through the full scientific workflow: framing falsifiable questions, surveying prior work, designing analyses, and pre-registering predictions before touching outcome data. Prevents p-hacking, HARKing, and data leakage by locking hypotheses and decision rules in advance, with a git-based freeze and audit script. Verifies every claim against freshly reproduced evidence and red-teams conclusions before reporting.

282|26

↗

npx skills add K-Dense-AI/science-superpowers --all -g -y

All Skills in This Repository (15)

Pure Emerald Level Indicators

📦 In Repo

K-Dense-AI

preregistering-analysis

Enforce timestamped preregistration of hypotheses and analysis plans before results.

executing-analysis

Execute frozen pre-registered analysis plans with deterministic steps and validation rules.

writing-science-skills

Create structured science skill documents with required frontmatter and sections.

using-science-superpowers

Evaluate and invoke the appropriate Science Superpowers skill before responding.

designing-the-analysis

Drafts a standardized, pre-registered analysis workflow with structured data-flow mapping and explicit inputs/outputs for approved research questions.

framing-research-questions

Converts vague research ideas into precise, falsifiable framing documents with hypotheses and success criteria.

investigating-anomalous-results

Diagnose root causes of anomalous data analysis results using a four-phase workflow.

subagent-driven-analysis

Dispatch a fresh subagent per preregistered analysis step with two-stage reviews.

requesting-red-team-review

Dispatch a red-team reviewer subagent to critique scientific conclusions.

surveying-prior-work

Compile established methods and prior effect sizes to ground analysis planning.

verifying-results-before-claiming

Verify research claims against reproduced outputs from raw data.

reporting-and-archiving-findings

Verify reproducibility, execute chosen archival actions, and archive code, data, and environment locks.

Official

Advanced

Frequently Asked Questions

FAQPage Schema

How to install Science Superpowers?▼

Run `npx skills add K-Dense-AI/science-superpowers --all -g -y` in your terminal to install all 16 research skills globally.

What does Science Superpowers do?▼

It gives your research agent a complete computational-science methodology: framing falsifiable questions, pre-registering predictions, running reproducible analyses, and verifying results before claiming them.

How does it prevent p-hacking and HARKing?▼

The pre-registration skill locks hypotheses, predictions, and decision rules before any outcome data is seen, and the included prereg.sh script proves from git history that predictions preceded results.

Does Science Superpowers work with Claude Code and Gemini CLI?▼

Yes. It runs on Claude Code, Gemini CLI, Codex, Copilot CLI, Cursor, Antigravity, and Pi, with tool-mapping references included for each platform.

Do I need statistics expertise to use it?▼

No. The skills trigger automatically when you ask your agent to investigate data, and they enforce sound methodology like power analysis, confound handling, and anomaly root-causing on your behalf.