·AgenticReadiness test site

AgenticReadiness test site

A deployable target for the AgenticReadiness Lumar custom metric. Every signal it detects is exposed as a correctly- and an incorrectly-implemented page, for end-to-end verification of a real crawl.

Site-wide signals are chosen by host. Page signals expose every variant as a route on any host. Files & live probes (robots.txt, /.well-known/*, sitemap, NLWeb, MCP auth) are one-per-host, so crawl the correct.* (all valid), broken.* (all present-but-wrong) and authbroken.* (valid except a broken MCP OAuth chain) hosts, plus the alt1…alt9 scenario hosts, which serve the alternative failure shapes no single artifact can hold at once. Locally, add ?mode=…, ?scenario=… or ?era=…. This host’s configuration is published at /__agentic-scenarios.json.

Composition pages

Every per-URL technology on one URL. Isolated fixtures prove each checker works; only these show whether they interfere — whether a page’s Product JSON-LD confuses the editorial-metadata reader, or one malformed block suppresses the technologies around it.

Discoverability

Can an agent find and enumerate the site — canonical URLs, Link headers, and the sitemap.

Content

Is the content legible to a no-JS agent — server rendering, a Markdown twin, JSON-LD, declared language and encoding.

Access

Does the site tell agents the rules — robots.txt, per-bot rules, Content-Signal, ai.txt, and TDMRep.

Capabilities

Can an agent act — WebMCP tools, schema:Action, an MCP server card, OpenAPI, NLWeb, and MCP authorization.

Commerce

Can an agent transact — Product/Offer schema and a Universal Commerce Protocol manifest.

Trust

Can an agent trust the source — editorial provenance, Web Bot Auth keys, and a signed A2A agent card.