History

We welcome the category we helped define.

Siteline began scanning websites for AI agent usability in March 2026, before the category had a name. This page is the receipts: every release date is recorded in our public changelog, and every external milestone links to a source you can check yourself. The source repository is private, so the first-commit hash below is provenance we state, not a link you can open.

Timeline

2026-03-19 — Siteline goes live. The repository starts fresh that day (first commit 26de9bc, “Initial AIReady scanner scaffold”) and version 1.0.0 ships the same day: 4-pillar SNAP scoring (Signal, Navigate, Absorb, Perform) and the public REST API (changelog).
2026-03-22 — Version 2.0.0: dynamic OG images, PDF export, MCP server, batch scanning, remediation tiers (changelog).
2026-03-26 — Version 2.1.0: the two-layer scoring model. SNAP fundamentals set the floor; Agentic Enablement caps the grade (changelog).
2026-04-17 — Cloudflare publishes isitagentready.com (blog published_at: 2026-04-17), bringing protocol-presence checks for agent-web standards to the category. A serious signal that the problem is real.
2026-04-27 — Paid services and platform launch (late April 2026): Audit, Site Ontology Workshop, contextual White-Label Reports, and Monitoring go live, alongside embeddable badges, a public benchmark corpus, and the SEO learn pages. The free scanner stays complete and ungated (changelog).
2026-05-18 — Versions 2.2.0–2.4.0: promise-relative Layer 2 with commerce-gated Transactability, Standards Library v1 (RFC 8288 Link headers, RFC 9727 api-catalog, OAuth discovery, MCP server-card, Web Bot Auth), and the F1 probe-evidence ledger (changelog).
2026-05-26 — Version 2.5.0: richer finding evidence, and static semantic-resource detection — assistant guides, ontology files, and relationship maps are reported as informational signals (changelog).
2026-05-30 — Version 2.6.0: security hardening release; SSRF validation covers IPv4-mapped private IPv6 (changelog).
2026-07-07 — Version 3.0.0: the composite scan. Every Siteline report now carries an attributed Cloudflare Agent Readiness baseline panel (informational, never feeds the grade) alongside Siteline’s SNAP quality judgment and open-standards conformance checks (Graceful Boundaries, GuideCheck). Grade-integrity fixes land in the same release (changelog).
2026-07-08 — Versions 3.1.0–3.1.1: the served changelog and API-manifest endpoints, OpenAPI documentation of the site-type parameter, and a portfolio-wide pricing and documentation reconciliation (changelog).

Category priority, and how we checked it

Siteline shipped on 2026-03-19. On 2026-08-03 we surveyed the public agent-readiness scanner field and checked each candidate against the Internet Archive, rather than trusting what any tool says about its own age. The finding: we could not find a working public agent-readiness scanner operating before Siteline. Here is what that rests on, including the two cases that came closest.

Cloudflare, 29 days later. isitagentready.com has no Internet Archive capture before 2026-04-17 — the same day as Cloudflare’s announcement, whose own structured data reads datePublished: 2026-04-17. Two independent sources agreeing.
Agent Checker: a waitlist, not a scanner. agentchecker.ai does have an earlier footprint — three captures, the only one with content dated 2025-10-10. But the archived page is a pre-launch email-capture page reading “Get Agent Ready” with a single address field. No URL input, no report, no scanner. Its own social-preview image is stamped 2025-09-30 and shows the same waitlist.
IsAgentReady: an uncorroborated date. isagentready.com declares datePublished: 2025-05-01 in its own structured data. The Internet Archive holds no captures of that domain at all, and we found no independent source for the date. A scanner running publicly since May 2025 would be expected to leave a trace. We report the claim; we could not verify it.
Old domains are not old tools. Several scanners in this field run on domains registered years earlier for unrelated businesses. A 2011 capture of a domain says nothing about when an agent-readiness tool appeared on it, and we did not count any such capture as priority.

We hold our own date to the same standard, and it fails the same test: the Internet Archive’s first capture of siteline.to is 2026-05-21, well after launch, and our repository is private, so 26de9bc on 2026-03-19 is provenance we assert rather than something you can independently confirm. The public changelog and this page are the audit trail we can offer. If you shipped an agent-readiness scanner before 2026-03-19, send us the evidence and this page changes.

The category today

Agent-readiness scanning now has real momentum, and that is good for everyone whose site needs to work for agents. Tools we track and respect:

Is It Agent Ready (Cloudflare) — protocol-presence checks across 21 standards in five categories. Its baseline now appears, attributed, inside every Siteline report.
Hard2Bit AI Scanner — agent-readiness scoring with a security angle.
SiteSpeakAI Agent Readiness Scanner — readiness checks oriented to chatbot and voice deployment.
Agent Checker — sends a live agent through a site and reports where it gets stuck.

Where Siteline differs is set out row by row in the comparison table below, including the rows where these tools lead.

Why it works the way it does

Two-layer scoring. Presence checks alone reward empty files. SNAP fundamentals set a floor from observed usability; Agentic Enablement caps the grade based on the quality of dedicated machine resources. The versioned rubric history is public in the API changelog.
Promise-relative judgment. A portfolio site and a store make different promises to an agent. Commerce sites carry a Transactability purpose at a higher bar; a static site is never penalized for not being a store.
Passive-only scanning. The scanner never triggers rate limits, never attempts transactions, and never sends credentials. What it cannot verify passively, it labels as passive evidence and says so.
Detection precision. A signal feeds the score only when it is strong enough to be trusted; everything else is reported as informational. Wrong grades are worse than missing checks.

How Siteline compares

Every competitor column below was read from that tool’s own live public pages on 2026-08-03 — what each says it checks, in its own words. This field moves fast and these tools ship often, so treat the columns as a dated snapshot and check the live scanners before relying on any row. Two of these rows do not favour us, and they are in the table for that reason.

Agent-readiness scanners compared on design decisions, not on which tool scores your site higher. Verified 2026-08-03.
Dimension	Siteline	Is It Agent Ready (Cloudflare)	Agent Ready (agent-ready.dev)	IsAgentReady
What a check earns	Quality, scored 0–2 per purpose. A resource must pass content validation; HTTP 200 alone earns nothing.	Presence across 21 standards in 5 categories.	Per-check pass/fail against a published spec.	Weighted categories; validates JSON-LD across 17 types.
Score architecture	Two independent layers. SNAP fundamentals set a floor; Agentic Enablement caps the grade. The lower one binds.	Single score across the five categories.	Per-check results plus a deterministic fix.	Single weighted score across five categories.
Site type	Changes the bar, not the weights. A commerce site carries an extra Transactability purpose; a static site is never penalised for not being a store.	Selects which checks run (content, site, API).	Uniform check set.	Uniform category weights.
Different content served to agents	Detected. If the agent-facing response drops most of the page’s substance, that caps the agentic level.	Not a listed check.	Not a listed check.	Not a listed check.
Per-probe evidence	Evidence ledger records each fetch and why it passed or failed validation.	Score plus generated recommendations.	Per-check pass/fail plus a deterministic fix.	Category findings.
Published check registry	No. The model, weights and grade bands are public in prose; the per-check registry is not.	Not published.	Yes — methodology and specs pages.	Category weights published; per-check registry not.
Ways to call it	Web and a free public REST API. Hosted API and remote MCP are contact-gated.	Web only.	Widest — web, REST, CLI, MCP, SDK and a browser extension.	Web only.
Open-standards conformance	Graceful Boundaries and GuideCheck conformance reported alongside the grade.	Protocol presence only.	Protocol presence, validated against specs.	Protocol presence only.
Cost of a full result	Free, complete, no account. Nothing is withheld behind a signup.	Free.	Free core scan; Pro tier for more.	Free.

The short version: presence checkers answer “does the file exist?” Siteline answers “does it actually help an agent, and does the site tell agents the same thing it tells people?” That is a deliberately narrower, harder question, and it is why a site can pass a presence scan and still not score well here.

Where others lead: Agent Ready publishes a fuller check registry than we do and is callable from more places; Cloudflare covers more emerging commerce and protocol standards by count. If a published per-check registry is what you need, use theirs — and tell us, because it is a fair thing to want from us too.