Annoying and alarming things about OpenCode (wren.wtf)

A recent critique has emerged against OpenCode, a highly popular open-source AI coding agent with over 161k stars on GitHub. The author expresses strong discontent with OpenCode’s security vulnerabilities and design failures, likening its impact to a "boot stamping on a human face." Key concerns include a poor security posture that could compromise user data and the design decisions leading to excessive computational inefficiencies. The codebase suffers from “prompt cache misses,” which can cause significant delays and context losses during tasks, particularly when the agent is interrupted or the context exceeds set limits. This is significant for the AI/ML community as it raises essential questions about the reliability and security of widely-utilized coding tools. The author highlights technical flaws such as unnecessary re-evaluations, inefficient context management, and user fatigue from decision prompts, suggesting that these issues undermine user productivity and the overall utility of OpenCode. The analysis calls for a reevaluation of methodologies employed in AI coding agents, emphasizing that responsiveness and security should be foundational, rather than afterthoughts. Such critiques serve as a cautionary tale, urging developers to prioritize robust, well-architected systems in AI tools.