docs: bump README to v3.1 stats (210 tasks; document risk_control / multi-site target_website)

README.md

@@ -26,30 +26,31 @@ LexBench-Browser is a benchmark for evaluating AI browser agents on real-world w
 
 ## Dataset Description
 
-This release contains **217 tasks** spanning common multi-step browsing flows. None of the tasks gate on user login, so they can be benchmarked end-to-end without manual sign-in.
+This release contains **210 tasks** spanning common multi-step browsing flows. None of the tasks gate on user login, so they can be benchmarked end-to-end without manual sign-in.
 
-### Dataset Statistics (v3.0, 2026-04-28)
+### Dataset Statistics (v3.1, 2026-04-29)
 
 | Attribute | Value |
 |-----------|-------|
-| Total tasks | 217 |
-| Version | 3.0 |
-| Languages | Chinese (141), English (76) |
+| Total tasks | 210 |
+| Version | 3.1 |
+| Languages | Chinese (137), English (73) |
+| `risk_control=true` tasks | 21 |
 
 #### Task type
 
 | Type | Count | Description |
 |------|-------|-------------|
-| T1   | 172   | Information retrieval |
-| T2   | 45    | Website operations |
+| T1   | 166   | Information retrieval |
+| T2   | 44    | Website operations |
 
 #### Reasoning type
 
 | Reasoning | Count |
 |-----------|-------|
-| single_step  | 122 |
-| multi_step   | 71  |
-| deep_analysis | 24 |
+| single_step  | 117 |
+| multi_step   | 70  |
+| deep_analysis | 23 |
 
 #### Domain Distribution
 
@@ -57,10 +58,10 @@ This release contains **217 tasks** spanning common multi-step browsing flows. N
 |--------|-------|
 | finance_gaming   | 44 |
 | video_platform   | 42 |
-| tools_education  | 41 |
+| tools_education  | 40 |
 | general          | 34 |
-| social_lifestyle | 31 |
-| ecommerce        | 24 |
+| social_lifestyle | 26 |
+| ecommerce        | 23 |
 | gaming           | 1  |
 
 ## Download
@@ -118,6 +119,8 @@ huggingface-cli download Lexmount/LexBench-Browser --repo-type dataset --local-d
   "difficulty": "medium",
   "login_required": false,
   "login_type": "",
+  "risk_control": false,
+  "risk_control_types": [],
   "target_website": "www.example.com",
   "language": "zh",
   "website_region": "zh",
@@ -144,9 +147,11 @@ huggingface-cli download Lexmount/LexBench-Browser --repo-type dataset --local-d
 - **language**: task description language (`zh` for Chinese, `en` for English).
 - **website_region**: target website region (`zh` for Chinese sites, `en` for international sites).
 - **login_required**: whether the task is login-gated (always `false` in this release).
+- **risk_control** / **risk_control_types**: whether the target site applies anti-crawl / risk-control mechanisms (CAPTCHA, slider verification, anti-bot, rate-limiting).
+- **target_website**: a single hostname, or for multi-site tasks, hostnames separated by `+` (e.g. `movie.douban.com + www.imdb.com`).
 - **robustness_tags**: ordered, deduplicated list across 6 categories / 16 tags (popup interference, sequence complexity, content dynamics, anti-crawl, localization, complex interaction).
 
-> **Migration note (v2.x → v3.0)**: the per-record `scenario_tier` field has been removed and the per-tier files (`l1.jsonl`, `l2.jsonl`, `l3-api.jsonl`, `l3-security.jsonl`, `tasks.jsonl`) have been replaced by a single `task.jsonl`. Slice the data with `login_required`, `domain`, or `risk_control` instead.
+> **Migration note (v2.x → v3.x)**: the per-record `scenario_tier` field has been removed and the per-tier files (`l1.jsonl`, `l2.jsonl`, `l3-api.jsonl`, `l3-security.jsonl`, `tasks.jsonl`) have been replaced by a single `task.jsonl`. Slice the data with `login_required`, `domain`, or `risk_control` instead.
 
 ## License