Internal Revenue Service | An official website of the United States government
AI Agentic Readiness
9 / 30
agentic checks passed 45% readiness - how well this site can be read, trusted and transacted with by AI agents
Last checked
irs.gov is live and passes 9 of 30 agentic-readiness checks (45%). It runs on drupal over HTTPS.
For AI agents, irs.gov's strongest signals are tidy, crawlable markup, fully indexable pages, a visually stable layout, and a clear canonical URL. Approximately 1300 DOM nodes - consider simplifying.
Its main gaps are no structured tables or lists, few clear definitions, no self-contained answer passages, and no Schema.org structured data. The weakest area is Structured data; addressing it first would raise this score the most.
Agent interaction
~Semantic HTML5 & AccessibilityOnly 72% of images have alt text
✓Clean, Crawlable HTMLPage has title and meta description
~Extraction Friction ScoreAverage sentence length: 27.9 words - slightly long
✓Indexability & Snippet EligibilityPages are indexable and snippet-eligible - the precondition for Google AI features is met
✓Layout Stability & Responsiveness17/18 images lack explicit width/height - the most common cause of layout shift (CLS)
✓Canonical URL StrategySingle canonical tag present (no duplicates)
✓Page Speed: Page SizeApproximately 1300 DOM nodes - consider simplifying
~Page Speed: Load Blockers3 render-blocking script(s) in <head>
Machine extraction
✗Table & List ExtractabilityNo HTML tables found across homepage
~Sentence AtomicityAverage 35% atomic sentences across 1 pages (target 40%+)
✓Entity DensityAverage entity density 35.4% across 1 pages (target >15%)
✗Definition PatternsOnly 1 definition pattern(s) found
✗Answer Capsule Pattern2/22 H2 sections have 15-30 word self-contained capsules (9%)
Structured data
✗Schema.org Structured DataNo JSON-LD structured data found across homepage
✗Speakable SchemaNo JSON-LD found across homepage - cannot assess speakable schema
AI discovery
✗llms.txt FileNo llms.txt file found at https://irs.gov/llms.txt (connection failed)
~robots.txt for AI CrawlersNo explicit AI crawler rules in robots.txt
✓Sitemap CompletenessSitemap index found, indicating organized sitemap structure
✗RSS/Atom FeedHomepage does not link to an RSS/Atom feed in <head>
✓Internationalization SignalsSelf-referencing hreflang present on homepage
✗Content Licensing & AI PermissionsNo ai.txt file found
Trust & transparency
~Entity Authority & E-E-A-TMultiple phone numbers found (3)
✗Creator TransparencyCreator visibility is weak on content-like pages
~Methodology TransparencyLittle content-production or review transparency is visible
~AI Assistance DisclosureNo AI-assistance disclosure found on sampled pages (neutral - the engine cannot tell whether AI was used)
✗Visible Date SignalNo visible date signals found across homepage (no <time> elements, no JSON-LD dates, no article meta dates)
✓Title & Meta Quality1/1 titles are shorter than 10 or longer than 70 characters
✗Content Freshness SignalsNo JSON-LD date properties (datePublished/dateModified) found across homepage
✗Author & Expert SchemaNo Person schema found across homepage
Every check is a verifiable fact about irs.gov's live pages. Scoring follows Google Lighthouse agentic-browsing: a fraction of concrete checks, not an invented 0-100.
How to improve for AI agents
The highest-impact changes to make irs.gov more agent-ready, in priority order.
1 Add structured tables and lists critical impactmedium effort
Present irs.gov's comparison and spec data as real HTML tables and lists agents can extract.
