Agent orchestration
Multi-agent research loops on Mac mini (seedmux coordinators, Claude Code, practice-oriented prompt packs) for literature, extraction, and nightly archives.
agent-tech practitioner · empirical economics
I study trust, reform, and civic capital in China with careful empirics — and I build the agent stacks that keep that research running.
Economist at Chongqing Technology and Business University (RePEc: pli1426). Training and collaborations span Chongqing University and applied micro topics on rural institutions, agricultural practices, and state-owned enterprise reform.
Day to day I pair R-based empirical workflows with multi-agent tooling — Mac-side orchestration, remote browser workers, and long-running research pipelines — so papers and systems improve each other.
Selected works (OpenAlex / SSRN / CEQ). Counts as of 2026-09.
Full citation graph via Google Scholar. Affiliation listing on IDEAS/RePEc.
Multi-agent research loops on Mac mini (seedmux coordinators, Claude Code, practice-oriented prompt packs) for literature, extraction, and nightly archives.
Steel Browser self-host on VPS for serial open → screenshot → pack pipelines; Mac keeps LLM extraction. CDP stays on localhost via SSH tunnel.
R on Apple Silicon for applied micro; Lexar disk for research archives; agent-readable llms.txt so crawlers and coding agents can navigate this site.
Small public MWEs on the same host — experiments, not the research CV.
Long-tail SEO niche site shell (guides, FAQ, command builder, ad placeholders).
Portal IA with category pages, /play SEO landings, and MIT embeds — aggregation, not hand-building every game.
Prefer academic or systems collaboration notes via Scholar / RePEc profiles above. This site is served from a personal VPS; infra notes for agents live in llms.txt.
Public HTTP only. Steel API / CDP are not exposed on this host.