CrawlDex
Check

Reports

Evidence reports for the agent-friendly web.

CrawlDex reports turn the live board into citable summaries with visible dates, method notes, and dispute paths.

Published

Current reports

Reviewed

Agent-Friendly Websites Report

A live report on which websites and task categories are easiest for AI agents to complete with visible evidence.

Last verified Jul 8, 2026. Based on 9,590 observations.

68
mean AES
21%
need a person nearby
Reviewed

State of the Agent Web Report

A launch report on where agents complete tasks, where they need a person nearby, and which blockers show up most often.

Last verified Jul 8, 2026. Based on 9,590 observations.

68
mean AES
21%
need a person nearby
Reviewed

Agent Web Index 2026-07

Reviewed monthly snapshots of public website-task outcomes, handoffs, blockers, and source-fingerprinted evidence.

Last verified Jul 8, 2026. Based on 9,590 observations.

68
mean AES
20%
need a person nearby
Review pending

The Agent Hostility Index

Website tasks where blockers, handoffs, and unstable browser surfaces are most likely to stop autonomous agents.

Last verified Jul 8, 2026. Based on 9,240 observations.

67
mean AES
22%
need a person nearby
Review pending

The Worst Subscription Cancellation Flows For AI Agents

Subscription cancellation routes ranked by blocker severity, expected handoff, retention friction, and low task reliability.

Last verified Jul 8, 2026. Based on 161 observations.

63
mean AES
52%
need a person nearby
Review pending

Which Travel Sites Are Safer For Agents?

Travel booking, cancellation, reservation, and lookup routes ranked by public preflight quality and agent-safe guardrails.

Last verified Jul 8, 2026. Based on 470 observations.

53
mean AES
35%
need a person nearby
Review pending

The Websites Most Likely To Break Autonomous Browsers

Routes where CAPTCHA, bot blocking, unstable DOMs, modals, or handoffs make autonomous browser execution fragile.

Last verified Jul 8, 2026. Based on 1,655 observations.

61
mean AES
72%
need a person nearby
Reviewed

The Return and Refund Policy Index For AI Agents

Return, refund, and eligibility routes ranked by observed policy clarity, support handoffs, blockers, confidence, freshness, and source caveats.

Last verified Jul 8, 2026. Based on 589 observations.

60
mean AES
35%
need a person nearby
Reviewed

The Checkout and Payment Friction Index For AI Agents

Checkout, booking, reservation, order, and payment routes ranked by observed user-present payment steps, hidden fees, account walls, and price ambiguity.

Last verified Jul 8, 2026. Based on 1,357 observations.

64
mean AES
52%
need a person nearby
Reviewed

Website Tasks That Require Login or 2FA For AI Agents

Routes where account login, two-factor authentication, or user-present identity steps limit autonomous browser agents.

Last verified Jul 8, 2026. Based on 2,315 observations.

76
mean AES
100%
need a person nearby
Reviewed

The Customer Support Handoff Index For AI Agents

Support and service routes where agents are most likely to encounter chat, phone, email, manual review, or other human handoffs.

Last verified Jul 8, 2026. Based on 1,780 observations.

63
mean AES
18%
need a person nearby
Reviewed

Why AI Agents Need A Human For Final Website Actions

Evidence-backed guide explaining where final confirmations, cancellations, and account-changing steps still need a person present.

Last verified Jul 8, 2026. Based on 6,116 observations.

77
mean AES
7%
need a person nearby
Reviewed

Why Login Walls Stop AI Agents

Evidence-backed guide to login, two-factor, and final-action authentication barriers across public CrawlDex rows.

Last verified Jul 8, 2026. Based on 5,974 observations.

77
mean AES
29%
need a person nearby
Reviewed

Why Payment Approval Blocks AI Agents

Evidence-backed guide to card approval, hidden-fee, price-mismatch, and payment-authorization boundaries for browser agents.

Last verified Jul 8, 2026. Based on 599 observations.

65
mean AES
93%
need a person nearby
Reviewed

Why Bot Blocking And 403s Stop AI Agents

Evidence-backed guide to access denial, HTTP 403, rate limits, bot blocking, and human verification outcomes in CrawlDex rows.

Last verified Jul 8, 2026. Based on 1,377 observations.

2
mean AES
5%
need a person nearby
Reviewed

Why Unstable Pages Confuse AI Agents

Evidence-backed guide to dynamic DOM changes, obstructing modals, sparse public text, and unclear policy surfaces.

Last verified Jul 8, 2026. Based on 1,785 observations.

70
mean AES
56%
need a person nearby
Reviewed

Where Human Handoffs Stop AI Agents

Evidence-backed guide to support, portal, and manual-review handoffs that move browser-agent work back to a person.

Last verified Jul 8, 2026. Based on 626 observations.

68
mean AES
98%
need a person nearby
Reviewed

Why Sparse Public Text Breaks AI Agent Evidence

Evidence-backed guide to sparse public copy, unclear policy text, and missing confirmations that weaken agent measurements.

Last verified Jul 8, 2026. Based on 1,040 observations.

66
mean AES
38%
need a person nearby
Reviewed

Why Popups And Modals Confuse AI Agents

Evidence-backed guide to overlays, popups, hidden controls, and modal interruptions in browser-agent workflows.

Last verified Jul 8, 2026. Based on 488 observations.

83
mean AES
77%
need a person nearby
Reviewed

Why Forced Account Creation Stops AI Agent Workflows

Evidence-backed guide to signup walls and account-creation boundaries that stop public browser-agent tasks.

Last verified Jul 8, 2026. Based on 268 observations.

73
mean AES
90%
need a person nearby
Reviewed

Why Fees And Price Mismatches Need User Review

Evidence-backed guide to observed hidden fees and price mismatches that require explicit user review.

Last verified Jul 8, 2026. Based on 510 observations.

63
mean AES
92%
need a person nearby
Reviewed

Why Return And Refund Flows Need Evidence Checks

Evidence-backed guide to return and refund routes with observed friction where agents need dated public evidence before acting.

Last verified Jul 8, 2026. Based on 466 observations.

59
mean AES
31%
need a person nearby
Reviewed

Why Rate Limits Stop AI Agent Measurement

Evidence-backed guide to HTTP 429 and rate-limit outcomes that stop repeatable browser-agent measurements.

Last verified Jul 8, 2026. Based on 58 observations.

2
mean AES
4%
need a person nearby
Reviewed

Why Regulated Sites Need User-Present Agent Mode

Evidence-backed guide to government, healthcare, insurance, legal, and finance workflows that require user-present boundaries.

Last verified Jul 8, 2026. Based on 1,492 observations.

78
mean AES
33%
need a person nearby
Reviewed

Why Retailer Order Flows Need Extra Agent Checks

Evidence-backed guide to retail order, checkout, pickup, delivery, return, and refund routes with observed friction.

Last verified Jul 8, 2026. Based on 724 observations.

59
mean AES
38%
need a person nearby
Reviewed

What AI Agents Should Preflight In API Docs

Evidence-backed guide to developer API documentation, rate-limit, OpenAPI, key, and webhook surfaces with observed friction.

Last verified Jul 8, 2026. Based on 1,918 observations.

73
mean AES
7%
need a person nearby