Summary
Christopher Burns, founder of Inth and creator of the open-source cookie-consent library C15T, recounts how his project went from 1,200 downloads at its Next.js Conf debut to 3 million NPM downloads with 45% month-on-month growth — and how Claude, ChatGPT, and Gemini became the number-one inbound discovery channel. The talk is a practical guide to optimizing developer tooling for the age of coding agents.
Burns frames the shift bluntly: software went from being installed by human wizards to being installed by agents responding to a prompt. That means the primitives of developer experience — documentation structure, discoverability, machine-readable formats — need to be rebuilt for an agent audience. His framework, which he has open-sourced as a tool called Lead Type, covers a layered set of optimizations: a hand-crafted LLM.txt (he found 40 well-written lines outperforms 1,000 auto-generated ones), twin .md versions of every documentation page, HTTP headers signaling markdown availability, and emerging Web MCP endpoints exposing search, page retrieval, and Q&A as tool calls.
The most counterintuitive finding: coding agents frequently never visit a library’s website at all — they read the node_modules directory and cached training data. Burns argues this makes keeping package-level documentation (READMEs, type signatures, inline comments) accurate and agent-readable as important as any website optimization, a perspective that applies equally to npm, PyPI, and Cargo ecosystems.
📺 Source: AI Engineer · Published August 26, 2026
🏷️ Format: Workflow Case Study
