Author(s): Youssef Hosni

Originally published on Towards AI.

Do AGENTS.md/CLAUDE.md Files Help Coding Agents? A New Paper Challenges this

Over the past year, adding an AGENTS.md or CLAUDE.md file to your coding repository has quietly become standard practice when using the coding agent. Agent vendors recommend it. Anthropic, OpenAI, and Qwen all encourage it. At the time of writing, over 60,000 public GitHub repositories already include one.

The research challenges the conventional wisdom that adding AGENTS.md or CLAUDE.md files boosts the efficiency of coding agents. Contrary to expectations, the results indicate that context files generated by LLMs often perform worse than having no context at all, highlighting a potential overload in information which can impair performance, especially in well-documented repositories. The study reveals that while human-written files demonstrate slight improvements, the cost of inference increases significantly, suggesting that agents may benefit from less redundancy in their instructions.

Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI