Tag
Argues that despite AI coding agents, humans still benefit from writing code for better understanding, ownership, and maintaining software quality.
Avi Press, founder of Scarf, explains the company's decision to move away from Haskell after seven years in production, driven by AI's transformation of development economics where slower compile times become a bottleneck for LLM-assisted coding.
The author reflects on how his own burner email blocklist blocked his legitimate alias, arguing that automatically blocking such domains at signup is counterproductive and recommends distinguishing between public shared inboxes and personal aliases.
A developer recounts experiencing burnout from constant interaction with LLMs, citing repetitive writing patterns and hallucinations as sources of fatigue.
Matt Pocock discovers it's trivial to build a proxy for reading raw system prompts sent to Anthropic from Claude Code, and he's now working to remove bloat in the system prompt.
A solo developer shares the pros and cons of building and maintaining Luxury Yacht, a desktop app for Kubernetes cluster management, emphasizing the freedom and responsibility of solo development.
A reflective blog post from 2010 examining the reasons why developers hate XML, tracing the hype and backlash cycles, and discussing the challenges of using XML for data modeling and interoperability.
After 3 years and 100k lines of Zig game logic code, using Zig alongside a C++ engine proved to be a lightweight and efficient development experience. Despite the language not yet being stable, the positive changes far outweigh the inconveniences.
A developer discusses the ongoing challenge of integrating AI agents into Slack and Microsoft Teams, noting that the integration layer—thread context, auth, streaming, permissions—often requires rebuilding despite improvements in agent logic tools like LangGraph and CrewAI.
Gergely Orosz argues that experienced software engineers remain valuable because they can discern good code from bad, something AI code generators still cannot reliably do.
The author shares the process of using AI tools (Claude, Codex, etc.) to build a landing page with a tech and ocean-floor aesthetic for the Lody AI product, emphasizing the development philosophy of humans controlling the lower limit and AI expanding the upper limit.
A distinguished engineer at AWS argues that AI tools have increased output volume but not genuine value, leading to stagnation in product quality and innovation.
User shares experience switching back from Codex to Claude Code, noting significant differences between the two.
A developer with 45 years of experience is building a local-first harness for LLMs with multi-agent logic, soon to be open-sourced on GitHub, and asks the community what features would improve their local LLM experience.
Guillermo Rauch praises Next.js's new 'Copy prompt' button for error messages, calling it a work of agentic art that improves developer experience.
OpenRouter introduces MCP integration, providing live model intelligence for agents to select and price models in real-time, simplifying long-running agent development.
A junior developer reflects on whether it is normal to find understanding and reviewing code changes more challenging than writing code when using AI coding tools.
Blog post discussing 'slop paralysis'—the inability to review code generated by AI coding agents—and offering strategies to mitigate it.
A developer shares that they spent 4 days using Codex to get an app running on Windows, finding the process absurd yet awesome.
A discussion about the manual effort required to gather context from Jira, Docs, and GitHub before running AI agents.