AERS collects 1,096 skills across 76 core skill sets for end-to-end social-science empirical research, with 9-stage automation and a pipeline harness. It includes 7 Stanford REAP × CoPaper.AI self-developed skills and is maintained by CoPaper.AI from Stanford REAP.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
Auto-Empirical Research Skills (AERS) is a repository that catalogs a large collection of agent skills for empirical research in social science. It presents 76 core skill bundles totaling 1,096 skills, organized into a 9-stage end-to-end research pipeline with a central orchestration tool.
How it works
The project segments workflows into nine stages, each with associated skills. It includes a project-wide orchestration component called Paper-WorkFlow that chains stages into an end-to-end pipeline. Skills span data retrieval, causal inference, estimation, robustness checks, table/figure production, writing, and post-AIGC publishing hygiene. The README lists a mix of self-developed and community skills, indicating vendor/hosted and community-contributed components integrated into the catalog.
Getting started
- The repository provides a centralized catalog at catalog/skills.json tracking all skills.
- It references a tour command: the 5-minute tour (make quickstart) and an IDE entry to initialize end-to-end execution via Paper-WorkFlow.
- The latest tagged release is v2026.07 (First tagged release on 2026-07-02) with a workflow to run validation checks and unit tests.
Excerpted commands and references from the README:
- "The 5-minute tour (
make quickstart) prints the same picture in your terminal." - Paper-WorkFlow is described as a one-click orchestrator: it can run the nine stages in sequence.
- Latest release details are provided under RELEASES (latest 1): v2026.07 AERS v2026.07 — first tagged release (2026-07-02): ## Highlights
- First tagged release. Every gate is green:
make check(validate + 183 unit tests + eval-harness + numeric benchmarks), validate-catalog, quality-evals, and OpenSSF Scorecard on
Recent releases
- v2026.07 AERS v2026.07 — first tagged release (2026-07-02): Highlights include green gates for make check, validate-catalog, quality-evals, and OpenSSF Scorecard.
Traction
- The repository has 3244 stars and 429 forks, with 0 open issues. Language: Stata. License: none listed. Created: 2026-04-03. Last push: 2026-08-04.
Behind the repo
- Collaboration appears to be between Stanford REAP and CoPaper.AI, as indicated by the README banner and trust surface sections.
Caveats
- License is listed as none; license status is not specified.
- Open issues: 0.
- The project uses a Chinese-English bilingual README structure; English entry points exist via README-en.md.