readiness-audit

name	readiness-audit
description	Route an ecommerce storefront through an Agentic Commerce readiness audit. Use when assessing whether a store should move to Custom Agent, Verified Audit, Forward Deployed Engineering, or Not Qualified.

Readiness Audit

Workflow

1. Confirm the submitted URL is a public production storefront.
2. Score each dimension below from public signals only, citing at least one URL or check per score.
3. Label evidence as public-signal only unless verified exports are explicitly provided.
4. Route using the thresholds, then state next steps and limitations.

Scoring rubric

Score each dimension 0-3.

Dimension	0	1	2	3
SEO basics	Not indexable, or robots/canonical broken	Indexable but thin metadata, weak internal links	Solid metadata and sitemap, minor gaps	Clean indexing, canonicals, sitemap, crawlable pages
AEO/GEO content	No answerable content	Some FAQ or policy text, buried or vague	Clear FAQs, policies, comparisons on stable URLs	Rich answer-shaped content with evidence and dates
Product knowledge	No structured data, sparse product copy	Partial attributes, no variants or availability	Product schema with price and availability, variant gaps	Complete attributes, variants, stock, offers in structured data
Agent access	Crawlers or AI bots blocked	Accessible but client-side-only rendering	Server-rendered content and sitemap, no agent affordances	Explicit affordances: llms.txt, feeds, or API/protocol paths
Policy/support	Policies missing or contradictory	Policies vague on returns, shipping, refunds	Clear policies and findable support channels	Policies precise enough for an agent to answer or act on
Commerce/action readiness	No bounded actions possible	Checkout works for humans only	Clear cart/checkout flow an agent could supervise	Protocol or API path toward agent-driven ordering

Routing thresholds

• Any dimension at 0, or total ≤ 6 → Not Qualified. Fix blocking gaps first; hand off to ai-search-remediation-plan.
• Total 7-11 → Verified Audit. Signals are mixed; verified exports (Search Console, analytics, order data) are needed before recommending a build.
• Total ≥ 12 with agent access ≥ 2 and policy/support ≥ 2 → Custom Agent. Hand off to custom-agent-remediation-plan.
• Total ≥ 12 but a standard setup cannot cover the platform, protocol, or integration gaps → Forward Deployed Engineering. Confirm with fde-opportunity-map.

A worked example with scores, evidence, and routing is in references/example-audit.md.

Guardrails

• Do not imply access to Search Console, analytics, revenue, logs, rankings, or conversions unless the user provided verified exports.
• See ../references/guardrails.md for shared cross-skill guardrails (autonomous action safety, evidence provenance, internal runtime disclosure).
• Prefer practical remediation over generic score commentary.