Auto-Empirical Research Skills
实证社科研究的 Agent 技能库 —— 全部 vendored、许可审计、安全扫描，并配备可执行的严谨性层：
数值恢复基准、方法陷阱评测、已发表论文复现 gold。
Vendored agent skills for empirical social science, with an executable rigor layer:
numeric recovery benchmarks, behavioral evals, and published-replication golds.
🔎 Skill Search · 技能搜索
Full-text + facet search over every cataloged skill: method, stage, language, license, commercial use.
docs/search.html →🧰 Tools Search · 工具搜索
Causal/econometrics libraries, autonomous research agents, MCP servers, causal discovery, datasets.
docs/tools-search.html →🛡️ Rigor Coverage · 严谨性覆盖
16 method families, each closed by an eval scenario and a deterministic numeric benchmark.
docs/RIGOR_COVERAGE.md →🪶 Benchmark Scoreboard · 公开基准榜
Reference vs naive baseline on every shared exam task — the rigor-layer gap, scored.
docs/BENCHMARK_SCOREBOARD.md →⚖️ Grade Your Agent · 外部评分协议
Point any agent (or a human with Stata) at the benchmark tasks and get an apples-to-apples score.
docs/INTEROP.md →📦 Repository · 仓库
README (中文默认 / English), install guide, plugin marketplace, releases.
github.com →💬 Discussions · 社区
Q&A, show-and-tell, and skill submissions. New here? Start with the welcome thread.
discussions →