Which agent-readiness tool should our team use?

The pick: Ora

Ora is the scan we recommend to a leadership team. It leads on total points, 94 to 72, but is behind on one criterion: tied to open standards, where 2 scanners score higher (3 for Ora, 5 for Cloudflare). How much that matters depends on how a reader weighs the criteria. The reasons, each with its source:

• 5 Nov 2026

  Ora's check catalogue listed 132 checks on 5 Nov 2026 in four layers: Discovery 16, Access 42, Usability 68 and Payments 6. 106 are verified and count toward the score; 26 are emerging and shown without scoring.

  Vendor claim. Ora check catalogue API, read 5 Nov 2026
• 5 Nov 2026

  Ora's llms.txt describes Journey: a real agent attempts a curated task against a domain, and partners holding an Ora-issued key can set a custom task instead.

  Vendor claim. Ora llms.txt, read 5 Nov 2026
• 5 Nov 2026

  Ora's home page reported 145,922 sites scanned on 5 Nov 2026.

  Vendor claim. Ora home page, read 5 Nov 2026
• 5 Nov 2026

  Ora's stats API counted 132,085 ranked companies on 5 Nov 2026, and its public leaderboard API listed 3,453 sites in 26 categories.

  Vendor claim. Ora stats API (GET /api/stats), read 5 Nov 2026
• 5 Nov 2026

  Ora's pricing page lists one plan, free, with unlimited score lookups and 10 scans a minute, and its llms.txt says no API key, credit card or signup is needed.

  Vendor claim. Ora pricing (JSON), read 5 Nov 2026
• 5 Nov 2026

  Ora's check catalogue lists every check with its layer and maturity, and 25 of the 132 also carry a link to the specification they test.

  Vendor claim. Ora check catalogue API, read 5 Nov 2026
• 5 Nov 2026

  Ora's methodology page sets out the layer weights of 20, 30, 40 and 10 points and an A+ to F grade scale.

  Vendor claim. Ora methodology, read 5 Nov 2026
• 19 Oct 2026

  Ora maintains the open AgentReady.org standard with Vercel, Stripe, Auth0, Mintlify, Tavily, Google, Elementor and MIT.

  Vendor claim. Ora blog, read 5 Nov 2026
• 7 Jul 2026

  Ora's blog dates three changes to how the score is computed: Deep Scan on 8 April 2026, Deep Scan version 1.1 on 13 May 2026 and version 2 on 7 July 2026.

  Vendor claim. Ora blog, read 5 Nov 2026

For a leader the practical result is three things on one page: a score and grade the board can track, a leaderboard to compare your sector against, and a test where an agent tries a real task. Your sector's leaderboard numbers are on their own page.

What Ora does not do

• It reads public pages only. Ora's llms.txt says it cannot scan private or internal sites or login-gated content.
• It publishes no data on repeat scans. Ora says the same scan run twice converges within a couple of points, which is a statement, not a dataset.
• Its fixing product is a waitlist. Ora's home page, read on 5 November 2026, says access is rolling out slowly.
• A score measures public pages. The ten decisions about bots, prices and payments stay with leadership, and a scan cannot make them.

The ten criteria in plain words

Whole journey, 15 points

Does the scan cover finding you, reading you, acting on your site and paying you?

Checks published, 15 points

How many checks does it publish?

Real agent test, 15 points

Does it send a real agent through a task?

Method in public, 10 points

Does it publish its check list, weights, grade scale and method changes?

Open standards, 5 points

Do its checks cite the open specifications they test?

Your team's tools, 10 points

Can your web team use it from an API, an MCP server, a command line or a CI gate?

Peer comparison, 10 points

Does it rank you against other sites?

Help to fix, 10 points

Does it say how to fix what failed, as a brief a coding agent can follow?

Repeat-scan evidence, 5 points

Does it publish evidence that scanning twice gives the same answer?

Free to try, 5 points

Can you try it free without signing up?

A point is awarded only where a tool's own pages state the capability, and every awarded point has its source under Sources. Every cell that first read as "not found" was read again on the vendor's own pages before this ranking was published, and a source found there raised that tool's score. Five more scanners were scored and rank below these: Mintlify (53), Lighthouse agentic browsing (39), SEOJuice (31), Microsoft Scope (27) and SiteSpeakAI (18).

Best for another job

StartupHub (startuphub.ai), 72 of 100

The cheapest scan by API, 0.001 USDC a call over x402 with no key, and a Live Agent Run on a free account.

Is Agentic (is-agentic.com), 65 of 100

Ora's checks grouped as Essential, Recommended and Bonus for public websites, free, with a CLI and a prompt to fix.

Agent Ready (agent-ready.dev), 61 of 100

A pull-request gate on its Pro plan where 69 of its 72 checks map to a published spec, and a plain pass percentage.

Cloudflare (isitagentready.com), 56 of 100

A free level read (0 to 5) with no account, and a keyless API and MCP server.

How a leader runs it

1. Enter your domain at Ora, or ask the web team to run npx ax@0.10 audit with your address.
2. Write down the score, the grade and the four layer scores.
3. Read the score against your sector's median on the sector guide.
4. Send the failed checks, ranked by impact, to the web team with an owner and a date.
5. Rescan after each fix and report the change in the quarterly board brief.

• Confirm your self-check with a scanWhat the four layers test and what a scan does to your site.
• What to ask a vendorThe same five questions put to Ora and three other kinds of supplier.