What does an HonestClaw scan show?
TL;DR
One HonestClaw scan of a public site returns four readings: Score, Dimensions, Checks, and Agent Comprehension.
- Score: a 0 to 100 Agent Readiness Score for how ready that public site is for agents to find, understand, and use what it publishes.
- Dimensions: which of nine live areas help agents and which hold them back.
- Checks: which scored findings drive the number, including what needs a fix.
- Agent Comprehension: whether agents can pull clear, consistent facts from the published pages the scan read. Not scored.
To see a current result, run a scan.
An HonestClaw scan shows what agents can get from one public site right now: the Agent Readiness Score, Dimensions, Checks, and Agent Comprehension. You get a 0 to 100 score for how ready that site is for agents to find, understand, and use what it publishes, which areas help or hold agents back, which scored findings drive the number, including what needs a fix, and whether clear, consistent facts come through on the pages the scan read.
The Agent Readiness Score
The Agent Readiness Score shows how ready one public site is for agents to find, understand, and use what that site publishes. It is a 0 to 100 summary of scored checks on those public surfaces. For the full product definition, see What is agent readiness?.
This scan only measures what the site publishes for anyone to read, so it does not log in, bypass access controls, or run live product tasks. It fetches pages, robots.txt, sitemaps, well-known endpoints, and linked documentation.
A higher score means more of the scored checks that help agents passed. A lower score means fewer of those scored checks passed.
On the public result, the score appears with its current band, such as Developing, and a confidence line, such as Confidence High. The same block shows the number of conclusive checks, evidence coverage, the scan time, and the scanner version.
The score overview lists nine areas, with each area's points next to its live name:
- Discoverability
- Content Accessibility
- Bot Access Control
- Protocol & Capability Discovery
- Authentication & Credentials
- Documentation Quality
- Structured Data I/O
- Observable State & Reliability
- Agent Safety & Trust
The free public scan reads a page as the server sends it, without running JavaScript. Most major AI crawlers do the same on first contact, so this is close to what they see. If a capability exists only after login, it sits outside this measurement.
Free and paid scans use the same scoring engine.
Dimensions
Dimensions breaks the Agent Readiness Score into nine areas so you can see which parts of readiness help agents and which hold them back. Each area shows how its points landed next to its live name: Discoverability, Content Accessibility, Bot Access Control, Protocol & Capability Discovery, Authentication & Credentials, Documentation Quality, Structured Data I/O, Observable State & Reliability, and Agent Safety & Trust.
It does not replace the check detail under Checks.
Checks
Checks are the individual scored findings that drive the Agent Readiness Score: what passed and what needs a fix. Only scored checks with conclusive results change the total. Unverified checks could not be confirmed; they do not raise or lower the score. Extra observations can add context without changing the score.
The Checks block groups work into method areas. Those labels are separate from the nine score-overview names. On the public result they appear as:
- Discovery. Can agents find the right entry points?
- Access & rendering. Can agents read the content?
- Permissions & policy. What access does the publisher permit?
- Documentation & meaning. Can agents understand the documentation?
- Interfaces & data. Which interfaces can agents use?
- Authentication & credentials. Can agents authenticate and manage credentials?
- Reliability & recovery. Can agents handle changes and failures?
- Identity & safeguards. Can agents identify the publisher and its safeguards?
At a glance, Checks can highlight Needs a fix, Furthest along, Largest measured gap, and how many conclusive results were recorded. You can open a method area for the checks inside it.
This is the place to find what to fix next.
Agent Comprehension
Agent Comprehension shows whether agents can pull clear, consistent facts from published pages on a website. Denser, consistent facts are what answer engines can cite and comprehension measures extractability. It does not cause citations by itself.
It is labeled Informational on the public result pages because it does not change the Agent Readiness Score. The block records informational answers from pages the scan could read, along with the pages that shaped those answers.
Static questions shown on the public are:
- What does this business do?
- Who is it for?
You can view evidence for a recorded answer. The free public scan results says whether other pages support, contradict, dilute, or do not speak to the homepage-derived answer. The supporting text explains that the scan uses the homepage it fetched as the reference for what the company is saying about itself, then checks other pages against that answer.
A paid scan allows you to use a source-of-truth to see if your homepage and other pages support your targeted message. Learn more about a deep scan here.
Agent Comprehension also shows evidence tallies. They use these live labels:
- Supports: the page backs the answer
- Contradicts: the page works against the answer
- Dilutes: the page weakens or muddies the answer without fully contradicting it
- None: the page does not speak to the answer
Limitations and next step
A scan is a measurement of a public site at a point in time. If a deeper look is offered, paying buys depth of analysis. It never buys a better number or a better rank.
If a result looks wrong, write disputes@honestclaw.ai. We correct errors. We do not change a result because someone dislikes it.
To see a current result for a site, run a scan.