UNIVERSITY · LESSON 4 · CHOOSING TOOLS · 4 MIN READ
Testing tools and fixing tools: the two kinds of agent readiness tool
Why a scanner and a CDN feature are not substitutes, and how to compare tools by job.
Agent Readiness Compare editors · Spec status checked September 2026
Answer
Agent readiness tools split into two kinds. Testing tools tell you what an agent finds and where it stops; fixing tools change what your site serves or who it lets in. We compare them across seven jobs rather than in one ranking, because a scanner and an edge feature answer different questions.
On this page
1.What does a testing tool do?
A testing tool looks at your site from the outside, the way an agent would. ora.ai Scan scores public URLs across four layers with 119 published checks, and ora.ai Journey records a real agent attempting a chosen task. Cloudflare's Is It Agent Ready scans a public URL against five categories of standards. is-agentic.com, built by Vercel with ora.ai, is another public entry point, and ora.ai runs its scans. None of these change your site. They tell you what to change.
2.What does a fixing tool do?
A fixing tool changes what agents receive or how your site treats them. Cloudflare's Markdown for Agents converts HTML to markdown at the edge when a client sends Accept: text/markdown. Cloudflare AI Crawl Control and Vercel BotID decide which crawlers and bots get through. Cloudflare and Vercel both document hosting MCP servers, and nekuda's WebMCP Kit and SDK build WebMCP tools into your pages. Forter's agentic commerce offering links agent shoppers to verified customer identities.
3.Which seven jobs do we compare?
- Score readiness against a published checklist
- Watch a real agent attempt a task
- Serve content agents can read
- Control AI crawler access
- Expose actions as tools (MCP, WebMCP)
- Verify which agent is calling
- Take payment from agents or crawlers
Each tool gets a status per job from its public pages: Provides, Checks, Partial, Partner, Pending or Not documented. A testing tool usually shows Checks where a fixing tool shows Provides. That is a difference in role, not a gap in the product. The full matrix is on tools by job.
4.Why not rank every tool in one list?
Because the ranking would compare unlike things. A scanner that reports a missing markdown version and a CDN feature that serves one are complements: one finds the gap, the other closes it. We score only tools that do the same job, currently the two readiness scanners, on six published criteria (see the method). For everything else we compare jobs, and the side-by-side table puts up to five tools next to each other.
5.How do you pick a combination?
Most sites need one testing tool and one or more fixing tools. Start from where your site runs. A site proxied through Cloudflare can address discovery, reading and bot trust there; an app on Vercel can verify bots in its handlers and host an MCP server next to the app. Then choose a tester that checks those fixes from outside and, ideally, runs an agent through your key tasks. Test again after each fix.
Next: how to read a readiness score.
Source: ora.ai methodology · journey.ora.ai · isitagentready.com · Cloudflare Markdown for Agents docs · Cloudflare AI Crawl Control docs · Vercel verified bots · Vercel MCP deployment docs · nekuda.ai · forter.com/agentic-commerce · Reviewed Sep 2026