Perhaps not Boring Technology after all

2026-03-09 · Source: Simon Willison's Weblog · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Advanced, quick

Summary

A recurring concern regarding Large Language Models (LLMs) for programming is their potential to bias technology choices towards tools well-represented in their training data, hindering the adoption of newer or less common alternatives. While this was evident a few years ago with languages like Python and JavaScript yielding better results than less widely used ones, the author observes a shift with the latest models from November 2025. These new models, integrated into effective coding agent harnesses, demonstrate excellent performance even with brand new, custom tools. By providing documentation via prompts like "use uvx showboat --help / rodney --help / chartroom --help to learn about these tools," agents leverage extended context lengths to understand and work with unfamiliar codebases and libraries, iterating and testing their output to fill knowledge gaps. This suggests coding agents may not enforce a "Choose Boring Technology" approach as initially expected.

Key takeaway

For AI Engineers evaluating coding agent capabilities, you should recognize that modern LLM-powered agents, particularly those with extended context windows, can effectively work with novel or proprietary technologies. Your teams can overcome potential biases towards widely used tools by explicitly feeding agents documentation or integrating official "skills" for less common libraries, enabling broader technology adoption without sacrificing agent utility.

Key insights

Modern coding agents can effectively utilize novel or private tools by consuming documentation and iterating.

Principles

• Longer context windows enhance agent adaptability.
• Agents can learn from existing code examples.

Method

Prompt agents with tool documentation (e.g., `--help` output) to enable understanding of new tools.

In practice

• Provide custom tool documentation to coding agents.
• Integrate agent skills for specific libraries/platforms.

Topics

• Large Language Models
• Coding Agents
• Context Window
• Technology Adoption
• Agent Skills

Code references

• remotion-dev/skills
• supabase/agent-skills
• vercel-labs/agent-skills
• prisma/skills

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

Related on AIssential

• My cheatsheet for a clean context — Effective coding agent use requires mindful context management, precise prompting, and systematic debugging.
• AI agents aren't replacing software engineering but expanding it far beyond code, researchers argue — AI agents expand software engineering beyond code, necessitating new methods for "semi-executable artifacts."
• TAI #186: Claude Code and the Christmas Awakening: Why CLI Agents Are Winning the Agentic Race — Agentic AI tools with file system access are driving significant productivity gains for both technical and non-technical users.
• Components of a Coding Agent: How LLMs Become Software Engineers — Coding agents transform LLMs into dynamic, multi-step problem solvers through structured system design.
• How I code with agents, without being 'technical' — AI agents enable non-technical users to build complex software by abstracting traditional coding into a guided, iterative process.
• Peter Yang on Small Teams, Coding Agents, and Why Human Ambition Has No Ceiling — AI agents are transforming software interaction and development, enabling radically smaller teams and automating knowledge work through conversational interfaces.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Simon Willison's Weblog.