Agent-Ready Docs — Daniel Kim

Making DoveRunner's developer documentation usable by AI coding agents, not just people — a docs audit, agent-ready site infrastructure, and an open-source tool that turns the method into something any team can run.

Developers increasingly meet our documentation through an AI coding agent, not a browser tab. This work asks a plain question — can an agent actually use these docs to ship a working integration? — and then fixes what the answer exposes.

Measure. Scanned the docs site with public agent-readiness scanners and audited the content against end-to-end customer integration scenarios, where gaps hide between pages.
Fix the infrastructure. llms.txt, a markdown copy of every page, markdown content negotiation, and explicit content signals for AI crawlers.
Fix the content. Worked through the audit findings with the product teams that own the answers.
Generalize it. Extracted the scan → fix → rescan loop into siteready, an open-source CLI and agent skill. This site is one of its test targets.

The same thread runs into everyday AI-assisted work: I used Claude with the Jira MCP to get up to speed as a new product owner, turning scattered tickets into usable context (write-up on Medium).