Meaning is not carried only by images
AGENTIC_NO_TEXT_AS_IMAGE
Images whose filename suggests they carry words - text, headline, quote, copy. Text baked into an image cannot be read, quoted or acted on: an agent asked to find your pricing will not find it if the price is a JPEG. Matched on filename only, deliberately: an empty alt attribute is the CORRECT way to mark a decorative image, so counting those would flag every properly-marked illustration on the page.
- Free, no signupNo account, no card
- No AI in the score986 deterministic rules
- Nothing publishedYour scans stay yours
- Answers in secondsQuick scan, no browser
How this rule is weighted
- Importance
- 5 / 10
- How much this matters relative to other rules.
- Confidence
- RESEARCH
- Backed by large-sample study or industry research, not a specification.
- Severity
- warning
- How the finding is presented when it fails.
- Scoring
- Counts
- A failure costs points in its category.
Impact is importance multiplied by the confidence tier’s weight — SPECIFICATION 1.0, RESEARCH 0.7, EMERGING 0.3, EXPERIMENTAL 0.1. Two rules backed by the same class of evidence therefore always carry the same weight, which is what makes the score reproducible rather than hand-tuned.
Questions about this rule
- Does AGENTIC_NO_TEXT_AS_IMAGE affect my score?
- Yes. It is an importance 5 rule at RESEARCH confidence, so a failure costs 5 × 0.7 points of impact in the AI Agent Readiness category before that category is normalised to 100.
- What does RESEARCH confidence mean?
- Backed by large-sample study or industry research, not a specification. Confidence is a declared tier rather than a per-rule number, so every rule backed by the same class of evidence carries the same weight - which is what makes the score reproducible instead of hand-tuned.
- How do I check AGENTIC_NO_TEXT_AS_IMAGE on my own site?
- Paste your URL into the box above and it runs only this rule, usually in a couple of seconds. It is also included in the AI Agent Readiness checker and in a full scan.
Tags
- agentic
- perceive
- content
Related AI Agent Readiness rules
- Every control has an accessible nameAn agent selects a control by its name - 'click Add to cart'. A control with no name can only be described positionally, which breaks the moment your layout changes.AGENTIC_BUTTONS_HAVE_NAMES
- Form fields are labelledAn agent filling a form matches your labels to what it was asked to enter. An unlabelled field is one it must guess at - and a wrong guess in a checkout is worse than a refusal.AGENTIC_INPUTS_LABELLED
- Interactive controls are real buttonsAgents locate controls through the accessibility tree, where a <div class="btn"> is not a control at all. This is the single most common reason an agent gets stuck on an otherwise ordinary page.AGENTIC_REAL_BUTTONS
- Navigation uses real linksAn agent follows href attributes. A div with a click handler is invisible to it - and unreachable without executing your JavaScript and guessing what the handler does.AGENTIC_REAL_LINKS
- A main content region is declaredAn agent orienting on an unfamiliar page looks for the main landmark first. Without one it has to guess which of forty divs holds the content, and it frequently guesses the navigation.AGENTIC_MAIN_CONTENT_MARKED
- Click handlers are not attached to generic elementsA div with onclick has no role, no accessible name and no keyboard behaviour. An agent cannot find it, and a keyboard user cannot reach it.AGENTIC_NO_DIV_CLICK_HANDLERS