Phronix: The Capability-Centric Learning Paradigm for Agents
Source
arxiv.org
Author
Hariharan Devarajan, Neeraj Rajesh, Aniruddha Marathe, Nikoli Dryden, Kin Ng, Michela Taufer
Date
Why it matters
AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → harnesses usually rely on handwritten skills and memory, so agents rediscover solutions each run. This work stores validated capabilities from executions and reports 93-95% first-attempt success on recurring tasks.
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
MCP — The Model Context Protocol — an open standard that lets any AI assistant plug into any tool or data source without custom integration code.
articleSubagents vs Agent Skills: Executing Reusable Knowledge for Long-Horizon Agentic TasksWasu Top Piriyakulkij, Rachel Lawrence, Alicia Curth, Sushrut Karmalkar, Niranjani Prasad
articleHPCAgent-Bench: A Framework for Sound Benchmarking of Code-Optimizing Agents, Harnesses, and SkillsYakup Koray Budanaz, Lorenzo Paleari, Marcin Copik, Alexandru Calotoiu, Timo Schneider, Benjamin Chen Ming Choong, Shriram Chandran, Lars Hulsbergen, Levi Widmer, Torsten Hoefler
articleCortex: Measurement-Gated Evolution and Pruning of Coding-Agent Skills and RulesSamuele Giampieri