Pete Was Right...(Again)

2026-06-19 · Source: Theo - t3․gg · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Advanced, quick

Summary

A new approach to interacting with coding agents suggests moving away from direct user-written prompts towards designing self-prompting, autonomous agent loops. The author, initially skeptical, found that traditional direct prompting, even with methods like the "Ralph loop," often led to increased error rates despite producing interesting results. Inspired by "Pete," the author successfully implemented systems where agents review code, provide feedback, make adjustments, and trigger re-reviews independently. This includes using tools like the "Hermes agent" to deliver context. After exploring and shipping code with these agent-driven loops, the author concludes that the majority of agent runs should ideally operate without user-written prompts, significantly enhancing productivity.

Key takeaway

For AI Engineers or Software Engineers developing with coding agents, shift your focus from writing direct prompts to designing self-prompting agent loops. This approach, where agents review, adjust, and re-trigger code changes autonomously, can significantly reduce error rates and boost productivity, as demonstrated by practical application. Consider implementing agent-driven feedback cycles to enhance your development workflow and improve code quality.

Key insights

Coding agents are more effective when designed to self-prompt and manage review cycles, rather than relying on direct user prompts.

Principles

• Direct user prompting of coding agents can increase error rates.
• Agents can effectively review, adjust, and re-trigger code changes.
• Self-prompting loops enhance agent productivity.

Method

Design agent systems where agents review code, provide feedback, make adjustments, and initiate re-reviews autonomously, using tools like Hermes agent for context delivery.

In practice

• Implement agent-driven code review loops.
• Configure agents to generate their own prompts.
• Utilize Hermes agent for context delivery.

Topics

• Coding Agents
• Agentic Workflows
• Self-Prompting AI
• Code Review Automation
• AI Development Practices

Best for: AI Architect, Machine Learning Engineer, AI Engineer, Software Engineer, MLOps Engineer

Related on AIssential

• Loop Engineering — Design autonomous systems ("loops") to prompt coding agents, replacing direct human interaction for iterative task completion.
• Before You Hit Push, Review Your Own Code — Proactive self-review, augmented by AI, elevates engineering quality and team collaboration.
• What's good for human developers is good for AI agents (and vice versa) — Agents can autonomously generate code changes and pull requests by leveraging internal data and collaborative prompting.
• Prompt Loops, Not Individual Instructions — Agent loops enable models to self-generate code for sub-prompting, unlocking advanced, self-directed workflows.
• I guess we're writing loops now? — Agents can dynamically orchestrate complex, multi-stage coding tasks through self-prompting loops, automating entire development workflows.
• 10 Lessons for Agentic Coding — Agentic coding is transforming software development, requiring new approaches and skills.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Theo - t3․gg.