AgentDocScore: Score Docs for AI Agents

Score and fix your docs site so coding agents can actually read it

Problem

Coding agents increasingly read docs before humans do, but most docs sites are hard for them: client-rendered pages, no llms.txt, no markdown versions, soft 404s, pages too big for a context window. A developer on HN built a checker for this, showing it's a real and overlooked pain point.

Solution

A CLI tool or SaaS that takes a docs URL, crawls it, scores it against the Agent-Friendly Documentation Spec, and gives concrete fixes: generate llms.txt, output markdown versions, split oversized pages, fix soft 404s. Integrate into CI to auto-check on deploy and open fix PRs.

Users

Developer relations teams, technical writers, open-source maintainers, and docs leads at SaaS companies

MVP

Only scan a URL and score it against the open spec; output an actionable fix list; generate basic llms.txt and markdown conversion; provide an online score page as lead magnet; support a GitHub Action.

Revenue

Free basic scans; paid subscription for advanced features (continuous monitoring, auto-fix PRs, private repo integration) at $19-49/mo per docs site.

Why now

AI coding agents are rapidly being adopted, but docs sites haven't been optimized for them yet—a brief window. An indie hacker can implement the spec quickly and establish a standard.