[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-82216":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":13,"openIssues":13,"contributorsCount":14,"subscribersCount":14,"size":14,"stars1d":15,"stars7d":16,"stars30d":17,"stars90d":14,"forks30d":14,"starsTrendScore":18,"compositeScore":19,"rankGlobal":9,"rankLanguage":9,"license":20,"archived":21,"fork":21,"defaultBranch":22,"hasWiki":21,"hasPages":21,"topics":23,"createdAt":9,"pushedAt":9,"updatedAt":24,"readmeContent":25,"aiSummary":26,"trendingCount":14,"starSnapshotCount":14,"syncStatus":15,"lastSyncTime":27,"discoverSource":28},82216,"science-superpowers","K-Dense-AI\u002Fscience-superpowers","K-Dense-AI","Composable computational-science methodology skills for AI research agents — pre-registration over TDD. A science-domain reimplementation of Superpowers.",null,"Shell",202,17,1,0,2,18,58,12,62.57,"Other",false,"main",[],"2026-06-12 04:01:37","# Science Superpowers\n\n[![License: MIT](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-MIT-yellow.svg)](LICENSE)\n[![Skills](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FSkills-15-brightgreen.svg)](#whats-inside)\n[![Follow on X](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FFollow_on_X-%40k__dense__ai-000000?logo=x)](https:\u002F\u002Fx.com\u002Fk_dense_ai)\n[![LinkedIn](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLinkedIn-K--Dense_Inc.-0A66C2?logo=linkedin)](https:\u002F\u002Fwww.linkedin.com\u002Fcompany\u002Fk-dense-inc)\n[![YouTube](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FYouTube-K--Dense_Inc.-FF0000?logo=youtube)](https:\u002F\u002Fwww.youtube.com\u002F@K-Dense-Inc)\n\nScience Superpowers is a complete computational-science methodology for your research agents, built on a set of composable skills plus initial instructions that make sure your agent actually uses them. It has **zero third-party dependencies** — it runs with only your agent harness and a POSIX shell.\n\n> ⭐ **If Science Superpowers helps your research, please [star this repository](https:\u002F\u002Fgithub.com\u002FK-Dense-AI\u002Fscience-superpowers).** A star helps other scientists and engineers find the project and tells us the methodology is worth expanding.\n>\n> **Stay up to date:** Follow K-Dense on [X](https:\u002F\u002Fx.com\u002Fk_dense_ai), [LinkedIn](https:\u002F\u002Fwww.linkedin.com\u002Fcompany\u002Fk-dense-inc), and [YouTube](https:\u002F\u002Fwww.youtube.com\u002F@K-Dense-Inc) for new skills, release announcements, and research workflow demos.\n\nIt is a reimplementation of [Superpowers](https:\u002F\u002Fgithub.com\u002Fobra\u002Fsuperpowers) (a software-development methodology) for a different domain: doing science with data. The architecture is the same — skills that auto-trigger via a session-start bootstrap — but the workflow is the research lifecycle, and the central discipline is **pre-registration** instead of test-driven development.\n\n## Contents\n\n- [How it works](#how-it-works)\n- [The basic workflow](#the-basic-workflow)\n- [Example: what using it looks like](#example-what-using-it-looks-like)\n- [What's inside](#whats-inside)\n- [Philosophy](#philosophy)\n- [Installation](#installation)\n- [Contributing](#contributing)\n- [License](#license)\n- [Star history](#star-history)\n\n## How it works\n\nIt starts the moment you fire up your agent. As soon as it sees you're trying to investigate something, it *doesn't* jump straight into running code on your data. Instead it steps back and helps you turn a fuzzy interest into a precise, falsifiable question.\n\nOnce the question is clear, it grounds the work in prior literature and standard methods, designs the analysis, and **pre-registers** the hypotheses, predictions, and decision rules *before looking at the outcomes*. That separation — confirmatory vs. exploratory, predictions locked before data — is what protects the work from p-hacking and HARKing (hypothesizing after results are known).\n\nThen it executes the pre-registered plan in a reproducible workspace (pinned environment, fixed seeds, immutable raw data), investigates anomalies by root cause instead of quietly dropping inconvenient data, verifies every claim against fresh reproduced evidence, and red-teams the result before reporting it.\n\nBecause the skills trigger automatically, you don't need to do anything special. Your research agent just has Science Superpowers.\n\n## The basic workflow\n\n1. **framing-research-questions** — Activates before any analysis. Turns a rough interest into a precise, falsifiable question with hypotheses, the data needed, and what would count as an answer. Saves a question document.\n2. **surveying-prior-work** — Grounds the question and chosen methods in what's already known: standard methods, known confounds, prior effect sizes.\n3. **designing-the-analysis** — Breaks the work into bite-sized analysis steps with exact datasets, variables, models\u002Ftests, power, and decision rules.\n4. **preregistering-analysis** — The Iron Law. Locks hypotheses, directional predictions, and decision rules — and the confirmatory\u002Fexploratory split — before any outcome is seen.\n5. **setting-up-reproducible-analysis** — Isolated, reproducible workspace: pinned environment, fixed seeds, immutable raw data, clean baseline.\n6. **subagent-driven-analysis** or **executing-analysis** — Carries out the pre-registered plan with review checkpoints.\n7. **investigating-anomalous-results** — Activates when results look wrong. Root-cause investigation before any adjustment.\n8. **verifying-results-before-claiming** — Evidence before claims: re-run, check assumptions, robustness, reproduce.\n9. **requesting-red-team-review** \u002F **receiving-critical-review** — Adversarial review before you believe or report a result.\n10. **reporting-and-archiving-findings** — Reproducibility check, then write-up\u002Fpreprint\u002Fiterate\u002Fshelve\u002Fdiscard, then archive code + data + environment.\n\n**The agent checks for relevant skills before any task.** Mandatory workflows, not suggestions.\n\n## Example: what using it looks like\n\nYou don't invoke anything. You just talk to your agent like normal — the skills trigger themselves.\n\n> **You:** I have last quarter's churn data. Does our new onboarding flow reduce churn?\n\nA vanilla agent would load the CSV and start slicing. With Science Superpowers, the agent steps back first:\n\n> **Agent:** Before touching the data, let me frame this precisely (`framing-research-questions`). The question is whether the new onboarding flow *causes* lower 90-day churn — but users weren't randomly assigned to it, so cohort and seasonality are confounders. Here's the falsifiable version, the data we'd need, and what would count as an answer…\n>\n> *[surveys prior internal analyses and standard methods, then designs the analysis]*\n>\n> **Agent:** I'm now **pre-registering** the hypothesis, the directional prediction (new flow → lower churn), and the decision rule **before** I look at any outcome (`preregistering-analysis`). Everything beyond this that wasn't pre-registered will be labeled exploratory.\n>\n> *[runs the pre-registered plan in a pinned, seeded, reproducible workspace; an anomaly appears]*\n>\n> **Agent:** One cohort shows implausibly low churn. Rather than dropping it, I'm root-causing it (`investigating-anomalous-results`) — it turns out to be a logging gap, not a real effect.\n>\n> *[verifies every claim against freshly reproduced output, then red-teams its own conclusion before reporting]*\n\nThe result is an analysis where the confirmatory claims are protected from p-hacking and HARKing, the anomalies were explained rather than hidden, and every number can be reproduced.\n\n## What's inside\n\n### Skills library\n\n**Framing**\n- **[framing-research-questions](skills\u002Fframing-research-questions\u002FSKILL.md)** — Turn an interest into a falsifiable question (entry gate)\n- **[surveying-prior-work](skills\u002Fsurveying-prior-work\u002FSKILL.md)** — Ground the question and methods in existing literature\n\n**Planning & pre-registration**\n- **[designing-the-analysis](skills\u002Fdesigning-the-analysis\u002FSKILL.md)** — Detailed, bite-sized analysis plan\n- **[preregistering-analysis](skills\u002Fpreregistering-analysis\u002FSKILL.md)** — Lock predictions and decision rules before seeing outcomes (includes statistical-fallacies reference)\n\n**Execution**\n- **[subagent-driven-analysis](skills\u002Fsubagent-driven-analysis\u002FSKILL.md)** — Fresh subagent per analysis step with two-stage review\n- **[executing-analysis](skills\u002Fexecuting-analysis\u002FSKILL.md)** — Inline batch execution with checkpoints\n- **[dispatching-parallel-investigations](skills\u002Fdispatching-parallel-investigations\u002FSKILL.md)** — Concurrent independent investigations\n\n**Discipline**\n- **[investigating-anomalous-results](skills\u002Finvestigating-anomalous-results\u002FSKILL.md)** — 4-phase root-cause process for surprising results\n- **[verifying-results-before-claiming](skills\u002Fverifying-results-before-claiming\u002FSKILL.md)** — Evidence before claims\n\n**Review**\n- **[requesting-red-team-review](skills\u002Frequesting-red-team-review\u002FSKILL.md)** — Dispatch a skeptical reviewer to attack the analysis\n- **[receiving-critical-review](skills\u002Freceiving-critical-review\u002FSKILL.md)** — Respond to critique with rigor, not performative agreement\n\n**Workspace & reporting**\n- **[setting-up-reproducible-analysis](skills\u002Fsetting-up-reproducible-analysis\u002FSKILL.md)** — Isolated, reproducible workspace\n- **[reporting-and-archiving-findings](skills\u002Freporting-and-archiving-findings\u002FSKILL.md)** — Decide how to report; archive code, data, environment\n\n**Meta**\n- **[writing-science-skills](skills\u002Fwriting-science-skills\u002FSKILL.md)** — Create new skills following the testing methodology\n- **[using-science-superpowers](skills\u002Fusing-science-superpowers\u002FSKILL.md)** — Introduction to the skills system\n\n## Philosophy\n\n- **Pre-registration** — State predictions and decision rules before seeing outcomes\n- **Confirmatory vs. exploratory** — Always labeled, never blurred\n- **Reproducibility** — Pinned environments, fixed seeds, immutable raw data\n- **Evidence over claims** — Verify before declaring a finding\n- **Root cause over patching** — Investigate anomalies; don't quietly drop data\n\n## Installation\n\nInstallation differs by harness. If you use more than one, install Science Superpowers separately for each.\n\n### Cursor\n\nIn Cursor Agent chat, install from the plugin marketplace, or point Cursor at this repository as a plugin. The `sessionStart` hook (`hooks\u002Fhooks-cursor.json`) loads the bootstrap automatically.\n\n### Claude Code\n\nRegister a marketplace pointing at this repo (`.claude-plugin\u002Fmarketplace.json`) and install the `science-superpowers` plugin. The `SessionStart` hook (`hooks\u002Fhooks.json`) loads the bootstrap.\n\n### Codex\n\nUse the committed Codex manifest at `.codex-plugin\u002Fplugin.json`.\n\n### Gemini CLI\n\nInstall as an extension; `gemini-extension.json` points the context file at `GEMINI.md`, which loads the bootstrap and the Gemini tool mapping.\n\n### OpenCode\n\nSee [.opencode\u002FINSTALL.md](.opencode\u002FINSTALL.md).\n\n### Google Antigravity\n\nAntigravity natively supports Agent Skills (the same `SKILL.md` format) and reads `GEMINI.md` \u002F `AGENTS.md` \u002F `.agent\u002Frules\u002F` as always-on rules at session start. Install the skills and load the bootstrap rule — see [.antigravity\u002FINSTALL.md](.antigravity\u002FINSTALL.md).\n\n## Contributing\n\nSee `AGENTS.md` \u002F `CLAUDE.md` for contributor guidelines, and `skills\u002Fwriting-science-skills\u002FSKILL.md` for the complete guide to creating and testing skills.\n\n## License\n\nMIT License — see the LICENSE file. This project reimplements the architecture of [Superpowers](https:\u002F\u002Fgithub.com\u002Fobra\u002Fsuperpowers) by Jesse Vincent.\n\n## Star history\n\n\u003Ca href=\"https:\u002F\u002Fwww.star-history.com\u002F?repos=K-Dense-AI%2Fscience-superpowers&type=date&legend=top-left\">\n \u003Cpicture>\n   \u003Csource media=\"(prefers-color-scheme: dark)\" srcset=\"https:\u002F\u002Fapi.star-history.com\u002Fchart?repos=K-Dense-AI\u002Fscience-superpowers&type=date&theme=dark&legend=top-left\" \u002F>\n   \u003Csource media=\"(prefers-color-scheme: light)\" srcset=\"https:\u002F\u002Fapi.star-history.com\u002Fchart?repos=K-Dense-AI\u002Fscience-superpowers&type=date&legend=top-left\" \u002F>\n   \u003Cimg alt=\"Star History Chart\" src=\"https:\u002F\u002Fapi.star-history.com\u002Fchart?repos=K-Dense-AI\u002Fscience-superpowers&type=date&legend=top-left\" \u002F>\n \u003C\u002Fpicture>\n\u003C\u002Fa>\n","Science Superpowers 是一个专为AI研究代理设计的计算科学研究方法学工具，通过一系列可组合的技能和初始指令确保代理能够有效利用这些技能。其核心功能包括零第三方依赖、仅需POSIX shell环境即可运行，并且强调预注册机制以区分确认性与探索性分析，从而避免p-hacking等问题。该工具特别适合于需要进行数据驱动科学研究的场景，如实验设计、数据分析及结果验证等过程，帮助研究人员在项目初期就明确研究问题并制定详细的分析计划。","2026-06-11 04:08:05","CREATED_QUERY"]