Ask HN: What is your (AI) dev tech stack / workflow? (June 2026)
An experienced developer asked HN for their current AI-powered development tech stacks and workflows to inform an upcoming workshop. The community responded with a wealth of innovative, often custom-built, agent-driven setups. This thread serves as a valuable snapshot of how cutting-edge developers are integrating AI into their daily coding practices.
The Lowdown
A veteran developer, accustomed to agile and XP methodologies but new to heavy AI integration, posted an "Ask HN" seeking modern AI development tools and workflows. Their goal is to prepare for workshops targeting both motivated newbies and existing software engineers looking to adopt AI tools. The original poster outlined their current stack (Linux Mint, VSCodium, Python, AWS) and various upcoming use-cases, from static site generation to web app development and personal infrastructure tasks, emphasizing open-source values and long-term supportability.
Key aspects of the request included:
- Seeking modern tooling suggestions for a diverse audience, from novices to experienced developers.
- A background in traditional, quality-focused development (TDD, well-documented tools, agile, XP).
- Planned use cases ranging from simple websites to web APIs and personal automation.
- Current stack is Linux-centric, but workshop participants are expected to use MacBooks and Windows.
- Curiosity about AI "harnesses" and agents as starting points.
The discussion quickly revealed that the landscape of AI-driven development is highly dynamic and personalized. Many respondents shared intricate, often bespoke, systems for leveraging AI, highlighting both the power and the complexity of these new paradigms. The conversation provided a rich collection of tools, philosophies, and workflows, illustrating the rapid evolution in how developers are approaching code generation, testing, and deployment with AI assistance.
The Gossip
The Agentic Ascent
Many developers are embracing highly automated, agent-driven workflows. These range from single-agent setups integrated into IDEs to multi-agent 'factories' that manage the entire development lifecycle, including PRDs, task decomposition, code generation, and review. The focus is on leveraging AI to handle repetitive tasks, maintain context, and accelerate development, often requiring custom tooling and careful architectural considerations to manage complexity and conflicts.
Tooling Triumphs
The discussion highlights a fragmented but rapidly evolving ecosystem of AI-powered development tools. Claude Code (often with Claude Opus/Anthropic Enterprise) and OpenAI's Codex are frequently mentioned as leading solutions, integrated with popular editors like VSCode, VSCodium, Neovim, and Ghostty. Users also explore tools like OpenCode for multiple models and KiloCode for VSCode derivatives, and discuss using OpenRouter for flexible model access. There's a clear preference for subscription models over token-based pricing for active users, alongside advice for integrating these tools into robust development environments.
The Human Element & Hurdles
While many are adopting AI, a significant thread discusses the role of human developers, the learning curve, and even outright skepticism. Some express caution, noting that relying too heavily on AI might lead to 'vibe coding' messes or that business requirements often remain the primary bottleneck. For teaching, some argue against AI for beginners to ensure fundamental understanding. Security companies, for example, face strict AI policies, limiting adoption. Others emphasize the importance of 'human in the loop' and careful review, even when using advanced agents.