Reproduce “One Skill Too Many: How Co-Installed Skills Conflict in Coding Agents” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Chaoliang Yan, Zihao Xu, Yuekang Li, Shangzhi Xu, Yi Liu, Gelei Deng, Siqi Ma
The authors study how similar co-installed coding-agent skills affect installed-skill requirements. They compare task completion and fidelity across 312 confirmed pairs on three Claude models, with 6,368 runs overall. Adding a similar skill reduced use of the installed skill by 19.9 percentage points, while task completion did not drop; exclusive core-function fidelity fell 5.6 points in these experiments.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.