@hitendra1007/agent-ready
v0.1.1
Published
Is your website ready for AI agents? Find out — and fix it.
Maintainers
Readme
agent-ready
Is your website ready for AI agents? Find out — and fix it.
A CLI + hosted scanner that audits a website for "agent-readiness": can an AI agent (using WebMCP, llms.txt, structured data, accessible forms/APIs) actually understand and act on this site? Generates a score + a report + optionally scaffolds the missing pieces (llms.txt, WebMCP manifest, structured data).
Why
As AI agents (browser agents, shopping agents, research agents) start acting on the web on users' behalf, sites that are hard for agents to parse/act on will lose traffic and transactions — the same way sites that ignored mobile or SEO fell behind. There's no simple, trusted "agent-readiness" audit tool yet. This is the Lighthouse/SEO-audit equivalent for the agentic web.
v0 scope
- CLI:
npx @hitendra1007/agent-ready scan https://example.com(published as a scoped package — npm's anti-squatting policy blocks bothagentreadyandagent-readyunscoped as "too similar" to an existing package. Once installed, the command itself is stillagent-ready.) - Checks for:
llms.txtpresence/validity- Structured data (schema.org, JSON-LD)
- robots.txt agent-related directives
- Basic WebMCP manifest presence
- Form/action accessibility (labeled inputs, semantic HTML)
- Outputs a score (0-100) + a markdown report with fix suggestions.
Non-goals (v0)
- Not a full accessibility auditor (that's a mature space — axe-core already exists).
- Not making automated changes to a live site — read-only audit first.
Status
v0.1.0 published to npm as @hitendra1007/agent-ready. Landing page: https://agent-ready.hitendra-patel2986.workers.dev. See docs/ARCHITECTURE.md and docs/ROADMAP.md.
Stack
TypeScript, Node.js CLI, cheerio (HTML parsing), optional hosted API later.
License
MIT (see LICENSE)
