What problem does it solve? Web frontends are rarely designed for the two machine audiences that now read them: answer engines deciding whether to cite the content, and AI agents trying to operate the UI. This Skill audits source code to find exactly which lines block crawlers, hide content from fetchers, or make controls unidentifiable to agents. ## Core Features & Use Cases - Static source fingerprinting: A probe script reports render mode (SSR/SSG/SPA), head assembly, descriptor files (robots.txt, llms.txt, sitemap, AGENTS.md), JSON-LD usage, and accessibility-tree surface before any review begins. - Five parallel review lenses: Programmatic path, human handoff, crawler & discovery, extractability, and agent operability — each with dedicated criteria files covering 14 AI crawlers, JSON-LD, citability signals, accessible names, canvas gaps, and export affordances. - Four audit modes: Full repo sweep, PR review lens anchored to diff lines, standalone llms.txt review, and author-facing component guidance. - Use Case: Reviewing a PR that touches robots.txt and head metadata — run the probe, apply the review lens to the diff, and get findings anchored to path:line with severity, instead of generic SEO advice. ## Quick Start Ask the agent to audit this repository's frontend for AI agent and LLM crawler readiness using the agent-web-audit skill.
