chore(release): 1.5.9 — strip zero-width title unicode (#33) + screenshot --clip/element (#34)

#33: some sites prepend runs of ZWJ/word-joiner/invisible-times/BOM to
document.title (badging/anti-scrape); left in, they polluted 'tab list', broke
text matching, and wrecked column alignment. sanitize_title() now strips
zero-width/bidi-format chars at every title ingestion point + get_title().

#34: 'screenshot <selector>' (element capture) already worked but was
undocumented; added 'screenshot --clip x,y,w,h' for an explicit pixel region
(CDP captureScreenshot clip), documented both in --help. Verified live.

.github/workflows/release-binaries.yml

@@ -147,16 +147,31 @@ jobs:
           git fetch --tags --force --quiet origin 2>/dev/null || true
           TAG="${{ github.event.inputs.tag || github.ref_name }}"
           PREV="$(git describe --tags --abbrev=0 "${TAG}^" 2>/dev/null || true)"
+          RANGE="${TAG}"
+          [ -n "$PREV" ] && RANGE="${PREV}..${TAG}"
+          # Group commit subjects by conventional-commit type so the notes are
+          # scannable ("what's new / what's fixed") instead of a flat dev log.
+          LOG="$(git log "$RANGE" --no-merges --pretty='%s' | grep -v '^chore(release)' || true)"
+          # NOTE: the job runs under `bash -e`. grep returning 1 (no match) and
+          # the `[ -n "$body" ]` test returning 1 (empty section) must NOT abort
+          # the script — otherwise a release whose commit range lacks a whole
+          # category (e.g. only `feat`, no `fix`) dies before writing the closing
+          # heredoc delimiter and the whole release step fails. `|| true` +
+          # `return 0` keep section() always-succeeding.
+          section() { # $1=header  $2=grep-pattern
+            local body; body="$(printf '%s\n' "$LOG" | grep -E "$2" | sed 's/^/- /' || true)"
+            [ -n "$body" ] && printf '\n### %s\n%s\n' "$1" "$body"
+            return 0
+          }
           {
             echo "notes<<__NOTES_EOF__"
             echo "## What changed"
-            echo ""
+            section "✨ Features" '^feat'
+            section "🐛 Fixes" '^fix'
+            section "🔧 Other" '^(perf|refactor|docs|build|ci|test|style|revert)'
             if [ -n "$PREV" ]; then
-              git log "${PREV}..${TAG}" --no-merges --pretty='- %s' | grep -v '^- chore(release)' || true
               echo ""
               echo "**Full changelog**: https://github.com/${{ github.repository }}/compare/${PREV}...${TAG}"
-            else
-              git log "${TAG}" --no-merges --pretty='- %s' | grep -v '^- chore(release)' || true
             fi
             echo "__NOTES_EOF__"
           } >> "$GITHUB_OUTPUT"

cli/Cargo.lock

@@ -290,7 +290,7 @@ checksum = "613afe47fcd5fac7ccf1db93babcb082c5994d996f20b8b159f2ad1658eb5724"
 
 [[package]]
 name = "chrome-use"
-version = "1.3.0"
+version = "1.5.9"
 dependencies = [
  "aes",
  "aes-gcm",

cli/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "chrome-use"
-version = "1.3.0"
+version = "1.5.9"
 edition = "2021"
 description = "Fast browser automation CLI for AI agents"
 license = "Apache-2.0"

cli/build.rs

@@ -3,6 +3,23 @@ use std::env;
 use std::fs;
 use std::path::Path;
 
+/// Embed the version of the `ab-connect` extension this CLI ships alongside, so
+/// `doctor` can tell a connected extension "you're older than what this CLI
+/// expects, update it." Read from the extension manifest at build time so it
+/// stays in sync with whatever extension version is in the same checkout/release
+/// (the ext is on its own 0.4.x line, separate from the CLI version). Falls back
+/// to "unknown" if the manifest can't be read.
+fn embed_extension_version() {
+    let manifest = Path::new("../extensions/ab-connect/manifest.json");
+    println!("cargo:rerun-if-changed=../extensions/ab-connect/manifest.json");
+    let version = fs::read_to_string(manifest)
+        .ok()
+        .and_then(|s| serde_json::from_str::<serde_json::Value>(&s).ok())
+        .and_then(|v| v.get("version").and_then(|x| x.as_str()).map(String::from))
+        .unwrap_or_else(|| "unknown".to_string());
+    println!("cargo:rustc-env=AB_CONNECT_VERSION={}", version);
+}
+
 /// Ensure `packages/dashboard/out/` exists so `rust-embed` doesn't fail during
 /// Rust-only dev builds where the dashboard hasn't been built. The placeholder
 /// `index.html` is only written when the directory is completely absent.
@@ -20,6 +37,7 @@ fn ensure_dashboard_dir() {
 
 fn main() {
     ensure_dashboard_dir();
+    embed_extension_version();
 
     let protocol_dir = Path::new("cdp-protocol");
     let out_dir = env::var("OUT_DIR").unwrap();

cli/src/commands.rs

@@ -30,12 +30,65 @@ pub enum ParseError {
     InvalidSessionName { name: String },
 }
 
+/// Top-level commands an agent is likely to mistype, used for "did you mean"
+/// suggestions on an unknown command (issue #29). Not exhaustive — just the
+/// common verbs plus a few known wrong-guesses mapped to the real command.
+const KNOWN_COMMANDS: &[&str] = &[
+    "open", "navigate", "click", "fill", "type", "press", "snapshot", "screenshot", "eval", "get",
+    "text", "html", "frames", "find", "wait", "scroll", "hover", "select", "check", "uncheck",
+    "tab", "tabs", "close", "back", "forward", "reload", "sessions", "status", "daemon", "doctor",
+    "upgrade", "connect", "cookies", "mouse", "keyboard", "stream", "frame", "profiles", "title",
+    "url", "is", "drag", "dialog", "upload",
+];
+
+/// Levenshtein distance, capped — small inputs only (command names).
+fn edit_distance(a: &str, b: &str) -> usize {
+    let a: Vec<char> = a.chars().collect();
+    let b: Vec<char> = b.chars().collect();
+    let mut prev: Vec<usize> = (0..=b.len()).collect();
+    let mut curr = vec![0usize; b.len() + 1];
+    for (i, &ca) in a.iter().enumerate() {
+        curr[0] = i + 1;
+        for (j, &cb) in b.iter().enumerate() {
+            let cost = if ca == cb { 0 } else { 1 };
+            curr[j + 1] = (prev[j + 1] + 1).min(curr[j] + 1).min(prev[j] + cost);
+        }
+        std::mem::swap(&mut prev, &mut curr);
+    }
+    prev[b.len()]
+}
+
+/// Closest known command within a small edit distance, or a prefix/substring
+/// match — `None` if nothing is close enough to suggest confidently.
+fn nearest_command(input: &str) -> Option<String> {
+    let lower = input.to_lowercase();
+    // Exact prefix/substring hits first (e.g. "session" -> "sessions").
+    if let Some(c) = KNOWN_COMMANDS
+        .iter()
+        .find(|c| c.starts_with(&lower) || lower.starts_with(**c))
+    {
+        return Some(c.to_string());
+    }
+    // Tolerance scales with length: short words get distance 1, longer get 2.
+    let max_dist = if lower.len() <= 4 { 1 } else { 2 };
+    KNOWN_COMMANDS
+        .iter()
+        .map(|c| (*c, edit_distance(&lower, c)))
+        .filter(|(_, d)| *d <= max_dist)
+        .min_by_key(|(_, d)| *d)
+        .map(|(c, _)| c.to_string())
+}
+
 impl ParseError {
     pub fn format(&self) -> String {
         match self {
-            ParseError::UnknownCommand { command } => {
-                format!("Unknown command: {}", command)
-            }
+            ParseError::UnknownCommand { command } => match nearest_command(command) {
+                Some(suggestion) => format!(
+                    "Unknown command: {}\nDid you mean: chrome-use {}?",
+                    command, suggestion
+                ),
+                None => format!("Unknown command: {}", command),
+            },
             ParseError::UnknownSubcommand {
                 subcommand,
                 valid_options,
@@ -424,6 +477,9 @@ fn parse_command_inner(args: &[String], flags: &Flags) -> Result<Value, ParseErr
         // === Core Actions ===
         "click" => {
             let new_tab = rest.contains(&"--new-tab");
+            // `--follow`: if the click opens a new tab, switch the active tab to
+            // it (default reports the opened tab but stays put) (issue #24-A).
+            let follow = rest.contains(&"--follow");
             // Coordinate click as a first-class form (issue #8.4): when the only
             // handle is a pixel position, no element/selector is needed.
             //   click <x> <y>          e.g. click 449 320
@@ -432,23 +488,27 @@ fn parse_command_inner(args: &[String], flags: &Flags) -> Result<Value, ParseErr
             let coord_args: Vec<&str> = rest
                 .iter()
                 .copied()
-                .filter(|a| *a != "--new-tab" && *a != "--coords")
+                .filter(|a| !a.starts_with("--"))
                 .collect();
             if let Some((x, y)) = parse_coords(&coord_args) {
                 return Ok(json!({ "id": id, "action": "click", "x": x, "y": y }));
             }
             let sel = rest
                 .iter()
-                .find(|arg| **arg != "--new-tab")
+                .find(|arg| !arg.starts_with("--"))
                 .ok_or_else(|| ParseError::MissingArguments {
                     context: "click".to_string(),
-                    usage: "click <selector> | click <x> <y> | click --coords <x>,<y> [--new-tab]",
+                    usage:
+                        "click <selector> | click <x> <y> | click --coords <x>,<y> [--new-tab] [--follow]",
                 })?;
+            let mut cmd = json!({ "id": id, "action": "click", "selector": sel });
             if new_tab {
-                Ok(json!({ "id": id, "action": "click", "selector": sel, "newTab": true }))
-            } else {
-                Ok(json!({ "id": id, "action": "click", "selector": sel }))
+                cmd["newTab"] = json!(true);
             }
+            if follow {
+                cmd["follow"] = json!(true);
+            }
+            Ok(cmd)
         }
         "dblclick" => {
             let sel = rest.first().ok_or_else(|| ParseError::MissingArguments {
@@ -583,11 +643,27 @@ fn parse_command_inner(args: &[String], flags: &Flags) -> Result<Value, ParseErr
 
         // === Keyboard ===
         "press" | "key" => {
-            let key = rest.first().ok_or_else(|| ParseError::MissingArguments {
-                context: "press".to_string(),
-                usage: "press <key>",
+            let key = rest.iter().find(|a| !a.starts_with("--")).ok_or_else(|| {
+                ParseError::MissingArguments {
+                    context: "press".to_string(),
+                    usage: "press <key> [--hold <ms>]",
+                }
             })?;
-            Ok(json!({ "id": id, "action": "press", "key": key }))
+            let mut c = json!({ "id": id, "action": "press", "key": key });
+            // `--hold <ms>`: hold the key down for <ms> then release, timed inside
+            // the daemon (one round-trip, no shell-sleep jitter) — for games and
+            // hold-to-charge where keydown+sleep+keyup over 3 round-trips is too
+            // imprecise.
+            if let Some(i) = rest.iter().position(|a| *a == "--hold") {
+                let ms = rest.get(i + 1).and_then(|s| s.parse::<u64>().ok()).ok_or(
+                    ParseError::MissingArguments {
+                        context: "press --hold".to_string(),
+                        usage: "press <key> --hold <ms>",
+                    },
+                )?;
+                c["hold"] = json!(ms);
+            }
+            Ok(c)
         }
         "keydown" => {
             let key = rest.first().ok_or_else(|| ParseError::MissingArguments {
@@ -831,17 +907,37 @@ fn parse_command_inner(args: &[String], flags: &Flags) -> Result<Value, ParseErr
             // selector: @ref or CSS selector
             // path: file path (contains / or . or ends with known extension)
             let mut full_page = false;
-            let positional: Vec<&str> = rest
-                .iter()
-                .filter(|arg| match **arg {
-                    "--full" | "-f" => {
-                        full_page = true;
-                        false
+            let mut clip: Option<Value> = None;
+            let mut positional: Vec<&str> = Vec::new();
+            let mut i = 0;
+            while i < rest.len() {
+                match rest[i] {
+                    "--full" | "-f" => full_page = true,
+                    // `--clip x,y,w,h` captures a pixel region (issue #34).
+                    "--clip" => {
+                        let raw = rest.get(i + 1).ok_or_else(|| ParseError::MissingArguments {
+                            context: "screenshot --clip".to_string(),
+                            usage: "screenshot --clip <x,y,w,h> [path]",
+                        })?;
+                        let nums: Vec<f64> = raw
+                            .split(',')
+                            .filter_map(|n| n.trim().parse::<f64>().ok())
+                            .collect();
+                        if nums.len() != 4 {
+                            return Err(ParseError::InvalidValue {
+                                message: format!("--clip expects 'x,y,w,h' (4 numbers), got '{raw}'"),
+                                usage: "screenshot --clip <x,y,w,h> [path]",
+                            });
+                        }
+                        clip = Some(json!({
+                            "x": nums[0], "y": nums[1], "width": nums[2], "height": nums[3]
+                        }));
+                        i += 1;
                     }
-                    _ => true,
-                })
-                .copied()
-                .collect();
+                    other => positional.push(other),
+                }
+                i += 1;
+            }
             let (selector, path) = match (positional.first(), positional.get(1)) {
                 (Some(first), Some(second)) => {
                     // Two args: first is selector, second is path
@@ -872,6 +968,9 @@ fn parse_command_inner(args: &[String], flags: &Flags) -> Result<Value, ParseErr
                 "path": path, "selector": selector,
                 "fullPage": full_page, "annotate": flags.annotate
             });
+            if let Some(c) = clip {
+                cmd["clip"] = c;
+            }
             if let Some(ref fmt) = flags.screenshot_format {
                 cmd["format"] = json!(fmt);
             }
@@ -1002,8 +1101,31 @@ fn parse_command_inner(args: &[String], flags: &Flags) -> Result<Value, ParseErr
             Ok(json!({ "id": id, "action": "stealth_status" }))
         }
 
+        // `cf-status` — Cloudflare challenge/clearance preflight: is the page
+        // currently a CF challenge, and is there a still-valid cf_clearance (the
+        // HttpOnly persistence cookie)? Lets an agent SKIP re-solving when already
+        // cleared, and know when it must solve. Persistence optimization.
+        "cf-status" | "cf" | "cloudflare-status" | "clearance" => {
+            Ok(json!({ "id": id, "action": "cf_status" }))
+        }
+
         // === Close ===
-        "close" | "quit" | "exit" => Ok(json!({ "id": id, "action": "close" })),
+        "close" | "quit" | "exit" => {
+            // `close <tab>` closes only that tab (and the output says "Tab
+            // closed"); bare `close` closes the browser/session. `close --all` is
+            // intercepted earlier in the dispatcher. Previously `close t12` still
+            // ran a browser close and alarmingly printed "Browser closed" (#26).
+            if let Some(tab_ref) = rest.iter().find(|a| !a.starts_with("--")) {
+                Ok(json!({ "id": id, "action": "tab_close", "tabId": tab_ref }))
+            } else {
+                Ok(json!({ "id": id, "action": "close" }))
+            }
+        }
+
+        // The active tab's stable handle — `targetId` survives cross-process
+        // navigation and is reusable across sessions, so an agent can hold it
+        // instead of re-deriving "which tab is live" from `tabs` each step (#26).
+        "current" => Ok(json!({ "id": id, "action": "current" })),
 
         // === Inspect ===
         "inspect" => Ok(json!({ "id": id, "action": "inspect" })),
@@ -1258,6 +1380,11 @@ fn parse_command_inner(args: &[String], flags: &Flags) -> Result<Value, ParseErr
         // === Get ===
         "get" => parse_get(&rest, &id),
 
+        // List every frame the session can reach (top + same-process child
+        // frames + out-of-process iframes), with a text-length per frame so you
+        // can see where a listing's description actually lives (issue #27).
+        "frames" => Ok(json!({ "id": id, "action": "frames" })),
+
         // Top-level shortcuts for `get <x>` status reads — users naturally type
         // `chrome-use url` / `cdp-url` / `title` without the `get` prefix
         // (and expect `cdp-url`/`cdp_url` to work interchangeably).
@@ -1559,11 +1686,16 @@ fn parse_command_inner(args: &[String], flags: &Flags) -> Result<Value, ParseErr
                     }
                     Ok(cmd)
                 }
-                Some(tab_ref) => Ok(json!({
-                    "id": id,
-                    "action": "tab_switch",
-                    "tabId": tab_ref,
-                })),
+                Some(tab_ref) => {
+                    // `tab <ref> --activate` (alias `--front`) switches to the tab
+                    // AND raises it to the foreground — for handing a specific tab
+                    // to the human (SMS code, captcha) (issue #24-C).
+                    let mut cmd = json!({ "id": id, "action": "tab_switch", "tabId": tab_ref });
+                    if rest.iter().any(|a| *a == "--activate" || *a == "--front") {
+                        cmd["activate"] = json!(true);
+                    }
+                    Ok(cmd)
+                }
                 None => {
                     let mut cmd = json!({ "id": id, "action": "tab_list" });
                     if full {
@@ -2345,11 +2477,42 @@ fn parse_get(rest: &[&str], id: &str) -> Result<Value, ParseError> {
 
     match rest.first().copied() {
         Some("text") => {
-            let sel = rest.get(1).ok_or_else(|| ParseError::MissingArguments {
-                context: "get text".to_string(),
-                usage: "get text <selector>",
-            })?;
-            Ok(json!({ "id": id, "action": "gettext", "selector": sel }))
+            // `get text --all-frames` aggregates visible text across every
+            // frame, including out-of-process iframes invisible to the top
+            // document (issue #27). The selector is ignored in this mode.
+            let all_frames = rest[1..]
+                .iter()
+                .any(|a| matches!(*a, "--all-frames" | "--frames" | "-a"));
+            if all_frames {
+                return Ok(json!({ "id": id, "action": "gettext", "allFrames": true }));
+            }
+            // `get text --pierce` reads through CLOSED shadow DOM / child docs
+            // via the CDP DOM tree — content eval/innerText can't reach, e.g. an
+            // extension's injected panel in a closed shadow root (issue #30).
+            let pierce = rest[1..]
+                .iter()
+                .any(|a| matches!(*a, "--pierce" | "--shadow" | "--deep"));
+            if pierce {
+                return Ok(json!({ "id": id, "action": "gettext", "pierce": true }));
+            }
+            // `get text --main` returns the main-content region (readability),
+            // skipping header/nav/footer/sidebar boilerplate (issue #27).
+            let main = rest[1..]
+                .iter()
+                .any(|a| matches!(*a, "--main" | "--readable" | "-m"));
+            if main {
+                return Ok(json!({ "id": id, "action": "gettext", "main": true }));
+            }
+            // `get text` with no selector reads the WHOLE PAGE and now defaults
+            // to cross-frame aggregation, so an agent gets a page's iframed
+            // content (listing descriptions etc.) without having to know about
+            // `--all-frames` (#27). On a single-frame page this is identical to
+            // the old body read; multi-frame pages get the child frames too —
+            // a strict superset. An explicit selector stays element-scoped.
+            match rest.iter().skip(1).find(|a| !a.starts_with("--")).copied() {
+                Some(sel) => Ok(json!({ "id": id, "action": "gettext", "selector": sel })),
+                None => Ok(json!({ "id": id, "action": "gettext", "allFrames": true })),
+            }
         }
         Some("html") => {
             let sel = rest.get(1).ok_or_else(|| ParseError::MissingArguments {
@@ -3581,6 +3744,22 @@ mod tests {
         assert_eq!(cmd["url"], "https://example.com");
     }
 
+    #[test]
+    fn test_press_plain_and_hold() {
+        let cmd = parse_command(&args("press d"), &default_flags()).unwrap();
+        assert_eq!(cmd["action"], "press");
+        assert_eq!(cmd["key"], "d");
+        assert!(cmd.get("hold").is_none());
+
+        let held = parse_command(&args("press d --hold 800"), &default_flags()).unwrap();
+        assert_eq!(held["key"], "d");
+        assert_eq!(held["hold"], 800);
+
+        // Missing/invalid duration is an error, not a silent no-hold.
+        assert!(parse_command(&args("press d --hold"), &default_flags()).is_err());
+        assert!(parse_command(&args("press d --hold abc"), &default_flags()).is_err());
+    }
+
     #[test]
     fn test_navigate_reuse_tab_flag() {
         let cmd = parse_command(
@@ -3747,6 +3926,21 @@ mod tests {
         assert!(cmd.get("x").is_none());
     }
 
+    #[test]
+    fn test_click_follow_flag() {
+        // `--follow` sets the flag; the selector is still found even with the flag
+        // before it (issue #24-A).
+        let cmd = parse_command(&args("click @e5 --follow"), &default_flags()).unwrap();
+        assert_eq!(cmd["selector"], "@e5");
+        assert_eq!(cmd["follow"], true);
+        let cmd2 = parse_command(&args("click --follow @e5"), &default_flags()).unwrap();
+        assert_eq!(cmd2["selector"], "@e5");
+        assert_eq!(cmd2["follow"], true);
+        // Absent by default.
+        let plain = parse_command(&args("click @e5"), &default_flags()).unwrap();
+        assert!(plain.get("follow").is_none());
+    }
+
     #[test]
     fn test_tabs_alias_lists() {
         assert_eq!(
@@ -3955,6 +4149,28 @@ mod tests {
         assert_eq!(cmd["tabId"], "docs");
     }
 
+    #[test]
+    fn test_close_tab_vs_browser() {
+        // `close <tab>` closes that tab (says "Tab closed"); bare `close` closes
+        // the browser (#26).
+        let tab = parse_command(&args("close t12"), &default_flags()).unwrap();
+        assert_eq!(tab["action"], "tab_close");
+        assert_eq!(tab["tabId"], "t12");
+        let browser = parse_command(&args("close"), &default_flags()).unwrap();
+        assert_eq!(browser["action"], "close");
+        // `quit`/`exit` aliases still browser-close.
+        assert_eq!(
+            parse_command(&args("quit"), &default_flags()).unwrap()["action"],
+            "close"
+        );
+    }
+
+    #[test]
+    fn test_current_command() {
+        let cmd = parse_command(&args("current"), &default_flags()).unwrap();
+        assert_eq!(cmd["action"], "current");
+    }
+
     #[test]
     fn test_tab_sends_string_tab_id() {
         let cmd = parse_command(&args("tab t2"), &default_flags()).unwrap();
@@ -4134,6 +4350,21 @@ mod tests {
         assert_eq!(cmd["fullPage"], true);
     }
 
+    #[test]
+    fn test_screenshot_clip() {
+        // `--clip x,y,w,h` captures a pixel region (issue #34); the path still parses.
+        let cmd = parse_command(&args("screenshot --clip 10,20,200,40 out.png"), &default_flags())
+            .unwrap();
+        assert_eq!(cmd["action"], "screenshot");
+        assert_eq!(cmd["clip"]["x"], 10.0);
+        assert_eq!(cmd["clip"]["y"], 20.0);
+        assert_eq!(cmd["clip"]["width"], 200.0);
+        assert_eq!(cmd["clip"]["height"], 40.0);
+        assert_eq!(cmd["path"], "out.png");
+        // Bad clip is a clear error, not silent.
+        assert!(parse_command(&args("screenshot --clip 1,2,3"), &default_flags()).is_err());
+    }
+
     #[test]
     fn test_screenshot_with_ref() {
         let cmd = parse_command(&args("screenshot @e1"), &default_flags()).unwrap();
@@ -4654,12 +4885,96 @@ mod tests {
     }
 
     #[test]
-    fn test_get_text_missing_selector() {
-        let result = parse_command(&args("get text"), &default_flags());
-        assert!(result.is_err());
-        let err = result.unwrap_err();
-        assert!(matches!(err, ParseError::MissingArguments { .. }));
-        assert!(err.format().contains("get text"));
+    fn test_get_text_defaults_to_all_frames() {
+        // `get text` with no selector now reads the whole page across ALL frames
+        // by default (#27), so iframed content isn't silently missed. (Was: a
+        // top-frame `body` read, #24-D.)
+        let cmd = parse_command(&args("get text"), &default_flags()).unwrap();
+        assert_eq!(cmd["action"], "gettext");
+        assert_eq!(cmd["allFrames"], true);
+        assert!(cmd.get("selector").is_none());
+        // An explicit selector still wins and stays element-scoped.
+        let cmd2 = parse_command(&args("get text h1"), &default_flags()).unwrap();
+        assert_eq!(cmd2["selector"], "h1");
+        assert!(cmd2.get("allFrames").is_none());
+        // `text` top-level shortcut behaves the same.
+        let cmd3 = parse_command(&args("text"), &default_flags()).unwrap();
+        assert_eq!(cmd3["allFrames"], true);
+    }
+
+    #[test]
+    fn test_get_text_all_frames() {
+        // `--all-frames` switches to whole-page, cross-frame aggregation and
+        // drops the selector (issue #27).
+        for variant in ["get text --all-frames", "get text --frames", "text -a"] {
+            let cmd = parse_command(&args(variant), &default_flags()).unwrap();
+            assert_eq!(cmd["action"], "gettext", "{variant}");
+            assert_eq!(cmd["allFrames"], true, "{variant}");
+            assert!(cmd.get("selector").is_none(), "{variant}");
+        }
+        // A flag mixed with a selector still triggers all-frames.
+        let cmd = parse_command(&args("get text body --all-frames"), &default_flags()).unwrap();
+        assert_eq!(cmd["allFrames"], true);
+        // Without the flag, a leading flag-like token is skipped for the selector.
+        let cmd = parse_command(&args("get text main"), &default_flags()).unwrap();
+        assert_eq!(cmd["selector"], "main");
+        assert!(cmd.get("allFrames").is_none());
+    }
+
+    #[test]
+    fn test_frames_command() {
+        let cmd = parse_command(&args("frames"), &default_flags()).unwrap();
+        assert_eq!(cmd["action"], "frames");
+    }
+
+    #[test]
+    fn test_nearest_command_suggestions() {
+        assert_eq!(nearest_command("sesions").as_deref(), Some("sessions"));
+        assert_eq!(nearest_command("session").as_deref(), Some("sessions"));
+        assert_eq!(nearest_command("clik").as_deref(), Some("click"));
+        assert_eq!(nearest_command("screenshits").as_deref(), Some("screenshot"));
+        // Nonsense with no close match stays silent.
+        assert_eq!(nearest_command("xyzzy"), None);
+        // The unknown-command error embeds the suggestion.
+        let err = ParseError::UnknownCommand {
+            command: "sesions".to_string(),
+        };
+        assert!(err.format().contains("Did you mean: chrome-use sessions?"));
+    }
+
+    #[test]
+    fn test_get_text_main() {
+        for variant in ["get text --main", "get text --readable", "text -m"] {
+            let cmd = parse_command(&args(variant), &default_flags()).unwrap();
+            assert_eq!(cmd["action"], "gettext", "{variant}");
+            assert_eq!(cmd["main"], true, "{variant}");
+            assert!(cmd.get("selector").is_none(), "{variant}");
+        }
+    }
+
+    #[test]
+    fn test_get_text_pierce() {
+        for variant in ["get text --pierce", "get text --shadow", "text --deep"] {
+            let cmd = parse_command(&args(variant), &default_flags()).unwrap();
+            assert_eq!(cmd["action"], "gettext", "{variant}");
+            assert_eq!(cmd["pierce"], true, "{variant}");
+            assert!(cmd.get("selector").is_none(), "{variant}");
+        }
+    }
+
+    #[test]
+    fn test_tab_activate_flag() {
+        let plain = parse_command(&args("tab t3"), &default_flags()).unwrap();
+        assert_eq!(plain["action"], "tab_switch");
+        assert!(plain.get("activate").is_none());
+
+        let act = parse_command(&args("tab t3 --activate"), &default_flags()).unwrap();
+        assert_eq!(act["action"], "tab_switch");
+        assert_eq!(act["tabId"], "t3");
+        assert_eq!(act["activate"], true);
+        // `--front` alias.
+        let front = parse_command(&args("tab t3 --front"), &default_flags()).unwrap();
+        assert_eq!(front["activate"], true);
     }
 
     // === Protocol alignment tests ===

cli/src/connect.rs

@@ -449,6 +449,69 @@ pub fn relay_url() -> Option<String> {
     }
 }
 
+/// Append a one-line record of how a CDP connection was established, to
+/// `~/.chrome-use/connect-mode.log`. This is the smoking-gun detector for the
+/// "Allow remote debugging?" consent modal: that modal ONLY appears on a raw
+/// remote-debugging attach / a browser we launched with a debug port — NEVER on
+/// the extension relay. When the modal reappears, this log says which session
+/// took which path and when, so we can tell a code regression (`raw-port` /
+/// `launched` while the relay was up) from Chrome's own extension-debugger
+/// consent UX. Low volume (one line per connection); best-effort, never fails a
+/// connection.
+pub fn log_connect_mode(ws_url: &str, launched: bool, session: &str) {
+    let relay = relay_url();
+    let relay_up = relay.is_some();
+    let mode = if launched {
+        "launched(debug-port)"
+    } else if relay.as_deref() == Some(ws_url) {
+        "relay"
+    } else if ws_url.contains("127.0.0.1") || ws_url.contains("localhost") {
+        "raw-port-attach"
+    } else {
+        "remote-ws"
+    };
+    // A raw-port attach or a self-launch while the relay was available is the
+    // exact thing that pops the consent modal — flag it loudly in the line.
+    let suspect = (mode == "raw-port-attach" || launched) && relay_up;
+    let line = format!(
+        "session={session} mode={mode} relay_up={relay_up}{} ws={ws_url}\n",
+        if suspect { " CONSENT-MODAL-RISK" } else { "" }
+    );
+    if let Some(home) = dirs::home_dir() {
+        let path = home.join(".chrome-use").join("connect-mode.log");
+        use std::io::Write;
+        if let Ok(mut f) = std::fs::OpenOptions::new()
+            .create(true)
+            .append(true)
+            .open(&path)
+        {
+            let _ = f.write_all(line.as_bytes());
+        }
+    }
+}
+
+/// Sidecar recording the connected extension's version, written by the host when
+/// it receives the extension's `hello` (sibling of `relay-cdp-url`). Lets
+/// `doctor` surface which extension build is live without a CDP round-trip.
+fn relay_ext_version_path() -> PathBuf {
+    relay_url_path().with_file_name("relay-ext-version")
+}
+
+/// Version of the connected `ab-connect` extension, if the host learned it from
+/// the extension's `hello`. `None` when no extension has connected since the
+/// host started, or the extension predates version reporting.
+pub fn relay_ext_version() -> Option<String> {
+    let s = std::fs::read_to_string(relay_ext_version_path())
+        .ok()?
+        .trim()
+        .to_string();
+    if s.is_empty() {
+        None
+    } else {
+        Some(s)
+    }
+}
+
 /// Hidden `__nm-host` mode: launched by Chrome for the ab-connect extension.
 ///
 /// Bridges the extension (native-messaging stdio, envelope protocol) to a local
@@ -570,6 +633,15 @@ async fn nm_host_main() {
             Ok(v) => v,
             Err(_) => continue,
         };
+        // Extension version handshake: record it next to the relay URL so
+        // `doctor` can report which extension build is live (and whether it's
+        // behind). Best-effort; the message carries no CDP payload.
+        if v.get("method").and_then(|m| m.as_str()) == Some("hello") {
+            if let Some(ver) = v.get("version").and_then(|x| x.as_str()) {
+                let _ = std::fs::write(relay_ext_version_path(), ver);
+            }
+            continue;
+        }
         let outs = {
             let mut s = state.lock().await;
             s.handle_ext_message(&v, "")
@@ -602,6 +674,7 @@ async fn nm_host_main() {
     }
     nm_log("[nm-host] stdin EOF — Chrome closed the port");
     let _ = std::fs::remove_file(relay_url_path());
+    let _ = std::fs::remove_file(relay_ext_version_path());
 }
 
 #[allow(clippy::too_many_arguments)]

cli/src/doctor/mod.rs

@@ -18,6 +18,7 @@ mod launch;
 mod network;
 mod providers;
 mod security;
+mod versions;
 
 use serde_json::{json, Value};
 
@@ -97,6 +98,7 @@ pub fn run_doctor(opts: DoctorOptions) -> i32 {
     let mut fixed: Vec<String> = Vec::new();
 
     environment::check(&mut checks);
+    versions::check(&mut checks);
     chrome::check(&mut checks);
     daemon::check(&mut checks);
     config::check(&mut checks);

cli/src/doctor/versions.rs

@@ -0,0 +1,89 @@
+//! Version-coherence checks across all four moving parts: the CLI binary, the
+//! per-session daemons (covered by `daemon.rs`), the connected `ab-connect`
+//! extension, and the bundled skill. The extension was previously a black box —
+//! nothing reported which build was live — so a user could sit on an old
+//! extension with no signal. The extension now reports its version over the
+//! relay (`hello`), the host records it, and this surfaces it in one place.
+
+use super::{Check, Status};
+use crate::{connect, upgrade};
+
+pub(super) fn check(checks: &mut Vec<Check>) {
+    let category = "Versions";
+    let cli_version = env!("CARGO_PKG_VERSION");
+
+    // CLI — compare against the latest seen by the background update check.
+    match upgrade::cached_latest_version() {
+        Some(latest) if upgrade::version_is_newer(&latest, cli_version) => {
+            checks.push(
+                Check::new(
+                    "versions.cli",
+                    category,
+                    Status::Warn,
+                    format!("CLI {cli_version} (newer available: {latest})"),
+                )
+                .with_fix("chrome-use upgrade".to_string()),
+            );
+        }
+        _ => {
+            checks.push(Check::new(
+                "versions.cli",
+                category,
+                Status::Pass,
+                format!("CLI {cli_version}"),
+            ));
+        }
+    }
+
+    // Extension — the build this CLI shipped alongside (embedded at compile time
+    // from the extension manifest) is what we expect to be running.
+    let expected_ext = env!("AB_CONNECT_VERSION");
+    match connect::relay_ext_version() {
+        Some(ext) if upgrade::version_is_newer(expected_ext, &ext) => {
+            checks.push(
+                Check::new(
+                    "versions.extension",
+                    category,
+                    Status::Warn,
+                    format!("extension {ext} is behind the bundled {expected_ext}"),
+                )
+                .with_fix(
+                    "update ab-connect in Chrome: chrome://extensions \u{2192} reload \
+                     (or wait for the Web Store auto-update)"
+                        .to_string(),
+                ),
+            );
+        }
+        Some(ext) => {
+            checks.push(Check::new(
+                "versions.extension",
+                category,
+                Status::Pass,
+                format!("extension {ext}"),
+            ));
+        }
+        None => {
+            checks.push(Check::new(
+                "versions.extension",
+                category,
+                Status::Info,
+                format!(
+                    "extension not connected (or it predates version reporting — \
+                     expected {expected_ext})"
+                ),
+            ));
+        }
+    }
+
+    // Skill — ships inside the same release artifact as the binary, so it's
+    // version-locked here. Copies made elsewhere via `skills add` aren't.
+    checks.push(Check::new(
+        "versions.skill",
+        category,
+        Status::Info,
+        format!(
+            "skills bundled with this CLI ({cli_version}); copies made via `skills add` \
+             elsewhere may be stale — re-run to refresh"
+        ),
+    ));
+}

cli/src/main.rs

@@ -893,6 +893,14 @@ fn main() {
         return;
     }
 
+    // `sessions` is a natural top-level guess for "list my sessions" (the skill
+    // advertises sessions as a feature) — route it to the daemon inventory the
+    // same way `daemon status` does (issue #29).
+    if clean.first().map(|s| s.as_str()) == Some("sessions") {
+        run_daemon(&["sessions".to_string(), "status".to_string()], flags.json);
+        return;
+    }
+
     // Handle close --all: close all active sessions
     if matches!(
         clean.first().map(|s| s.as_str()),

cli/src/native/actions.rs

@@ -1332,6 +1332,7 @@ pub async fn execute_command(cmd: &Value, state: &mut DaemonState) -> Value {
         "uncheck" => handle_uncheck(cmd, state).await,
         "wait" => handle_wait(cmd, state).await,
         "gettext" => handle_gettext(cmd, state).await,
+        "frames" => handle_frames(cmd, state).await,
         "getattribute" => handle_getattribute(cmd, state).await,
         "isvisible" => handle_isvisible(cmd, state).await,
         "isenabled" => handle_isenabled(cmd, state).await,
@@ -1340,6 +1341,7 @@ pub async fn execute_command(cmd: &Value, state: &mut DaemonState) -> Value {
         "forward" => handle_forward(state).await,
         "reload" => handle_reload(state).await,
         "cookies_get" => handle_cookies_get(cmd, state).await,
+        "cf_status" => handle_cf_status(cmd, state).await,
         "cookies_set" => handle_cookies_set(cmd, state).await,
         "cookies_clear" => handle_cookies_clear(state).await,
         "storage_get" => handle_storage_get(cmd, state).await,
@@ -1395,6 +1397,7 @@ pub async fn execute_command(cmd: &Value, state: &mut DaemonState) -> Value {
         "count" => handle_count(cmd, state).await,
         "styles" => handle_styles(cmd, state).await,
         "bringtofront" => handle_bringtofront(state).await,
+        "current" => handle_current(state).await,
         "timezone" => handle_timezone(cmd, state).await,
         "locale" => handle_locale(cmd, state).await,
         "geolocation" => handle_geolocation(cmd, state).await,
@@ -2888,7 +2891,32 @@ async fn handle_snapshot(cmd: &Value, state: &mut DaemonState) -> Result<Value,
         })
         .collect();
 
-    Ok(json!({ "snapshot": tree, "origin": url, "refs": refs }))
+    let ref_count = refs.len();
+    let mut out = json!({ "snapshot": tree, "origin": url, "refs": refs });
+
+    // Canvas/WebGL apps (games, map/3D viewers, drawing tools) paint to a
+    // <canvas> and expose almost no accessibility tree, so `snapshot` comes back
+    // near-empty and agents get stuck looking for refs that will never exist
+    // (dogfood: the Dead Cell game). When the tree is sparse but a canvas
+    // dominates the viewport, tell them to switch to the screenshot-driven path.
+    if ref_count < 3 {
+        let canvas_js =
+            "(() => { const c = document.querySelector('canvas'); if (!c) return false; \
+                         const r = c.getBoundingClientRect(); \
+                         return r.width * r.height > innerWidth * innerHeight * 0.5; })()";
+        if let Ok(v) = mgr.evaluate(canvas_js, None).await {
+            if v.as_bool() == Some(true) {
+                out["note"] = json!(
+                    "This page renders to a <canvas> (game / WebGL / editor) and exposes almost no \
+                     accessibility tree — refs won't help. Use `screenshot` to see it, coordinate \
+                     `click <x> <y>` to interact, and `keydown`/`keyup`/`press` for keyboard \
+                     (hold-to-move: `keydown d` … `keyup d`)."
+                );
+            }
+        }
+    }
+
+    Ok(out)
 }
 
 /// Resolve a (possibly relative) saved-file path to an absolute one so the CLI
@@ -2972,6 +3000,14 @@ async fn handle_screenshot(cmd: &Value, state: &mut DaemonState) -> Result<Value
             .get("screenshotDir")
             .and_then(|v| v.as_str())
             .map(String::from),
+        clip: cmd.get("clip").and_then(|c| {
+            Some((
+                c.get("x")?.as_f64()?,
+                c.get("y")?.as_f64()?,
+                c.get("width")?.as_f64()?,
+                c.get("height")?.as_f64()?,
+            ))
+        }),
     };
 
     if annotate {
@@ -3091,6 +3127,15 @@ async fn handle_click(cmd: &Value, state: &mut DaemonState) -> Result<Value, Str
 
     let button = cmd.get("button").and_then(|v| v.as_str()).unwrap_or("left");
     let click_count = cmd.get("clickCount").and_then(|v| v.as_i64()).unwrap_or(1) as i32;
+    let follow = cmd.get("follow").and_then(|v| v.as_bool()).unwrap_or(false);
+
+    // Snapshot tracked targets so we can tell if this click opened a NEW tab
+    // (target=_blank link / window.open). On the relay the new tab is discovered
+    // passively and doesn't steal focus (#7/#8.1), so without surfacing it the
+    // post-click snapshot shows the OLD page and looks like the click failed
+    // (issue #24-A).
+    let before: std::collections::HashSet<String> =
+        mgr.pages_list().into_iter().map(|p| p.target_id).collect();
 
     interaction::click(
         &mgr.client,
@@ -3103,7 +3148,26 @@ async fn handle_click(cmd: &Value, state: &mut DaemonState) -> Result<Value, Str
     )
     .await?;
 
-    Ok(json!({ "clicked": selector }))
+    // Give a just-opened tab a moment to register, then look for it.
+    let mgr = state.browser.as_mut().ok_or("Browser not launched")?;
+    tokio::time::sleep(std::time::Duration::from_millis(150)).await;
+    let opened = mgr.adopt_newly_opened(&before).await;
+
+    let mut out = json!({ "clicked": selector });
+    if let Some(page) = opened {
+        let tab_id = super::browser::format_tab_id(page.tab_id);
+        out["openedTab"] = json!({ "tabId": tab_id, "url": page.url, "title": page.title });
+        // `--follow`: switch the active tab to the newly-opened one (default is
+        // to report it but stay put, so multi-tab flows aren't hijacked).
+        if follow {
+            state.ref_map.clear();
+            state.iframe_sessions.clear();
+            state.active_frame_id = None;
+            let _ = mgr.tab_switch_by_id(page.tab_id).await;
+            out["followed"] = json!(true);
+        }
+    }
+    Ok(out)
 }
 
 async fn handle_dblclick(cmd: &Value, state: &mut DaemonState) -> Result<Value, String> {
@@ -3309,6 +3373,16 @@ async fn handle_press(cmd: &Value, state: &mut DaemonState) -> Result<Value, Str
     // Parse modifier+key chords like "Control+a", "Shift+Enter", "Control+Shift+a"
     let (actual_key, modifiers) = parse_key_chord(key);
 
+    // `--hold <ms>`: keyDown, wait, keyUp — all inside the daemon so the hold
+    // duration is precise (no shell-sleep / round-trip jitter). For games
+    // (hold-to-move/charge) and any press-and-hold interaction.
+    if let Some(ms) = cmd.get("hold").and_then(|v| v.as_u64()) {
+        interaction::dispatch_single_key(&mgr.client, &session_id, &actual_key, "keyDown").await?;
+        tokio::time::sleep(std::time::Duration::from_millis(ms)).await;
+        interaction::dispatch_single_key(&mgr.client, &session_id, &actual_key, "keyUp").await?;
+        return Ok(json!({ "pressed": key, "heldMs": ms }));
+    }
+
     interaction::press_key_with_modifiers(&mgr.client, &session_id, &actual_key, modifiers).await?;
     Ok(json!({ "pressed": key }))
 }
@@ -3530,6 +3604,56 @@ async fn handle_wait(cmd: &Value, state: &mut DaemonState) -> Result<Value, Stri
 async fn handle_gettext(cmd: &Value, state: &mut DaemonState) -> Result<Value, String> {
     let mgr = state.browser.as_ref().ok_or("Browser not launched")?;
     let session_id = mgr.active_session_id()?.to_string();
+
+    // `get text --all-frames` aggregates visible text across every frame the
+    // session can reach — including out-of-process iframes that never show up
+    // in the top document (#27: Yahoo/Rakuten/Mercari listing descriptions).
+    if cmd.get("allFrames").and_then(|v| v.as_bool()) == Some(true) {
+        let frames = super::element::collect_all_frames_text(
+            &mgr.client,
+            &session_id,
+            &state.iframe_sessions,
+        )
+        .await?;
+        let mut combined = String::new();
+        let mut frame_count = 0usize;
+        for f in &frames {
+            let t = f.text.trim();
+            if t.is_empty() {
+                continue;
+            }
+            frame_count += 1;
+            if f.kind != "top" {
+                combined.push_str(&format!("\n\n----- frame [{}] {} -----\n", f.kind, f.url));
+            }
+            combined.push_str(t);
+        }
+        let url = mgr.get_url().await.unwrap_or_default();
+        return Ok(json!({
+            "text": combined,
+            "origin": url,
+            "frames": frame_count,
+            "allFrames": true,
+        }));
+    }
+
+    // `get text --pierce` reads text through CLOSED shadow roots and child
+    // documents via the CDP DOM tree — content `innerText`/`eval` can't see,
+    // e.g. an extension's injected panel in a closed shadow DOM (#30).
+    if cmd.get("pierce").and_then(|v| v.as_bool()) == Some(true) {
+        let text = super::element::get_pierced_text(&mgr.client, &session_id).await?;
+        let url = mgr.get_url().await.unwrap_or_default();
+        return Ok(json!({ "text": text, "origin": url, "pierce": true }));
+    }
+
+    // `get text --main` returns the page's main-content region (readability-lite),
+    // skipping global header/nav/footer/sidebar boilerplate (#27).
+    if cmd.get("main").and_then(|v| v.as_bool()) == Some(true) {
+        let text = super::element::get_main_content_text(&mgr.client, &session_id).await?;
+        let url = mgr.get_url().await.unwrap_or_default();
+        return Ok(json!({ "text": text, "origin": url, "main": true }));
+    }
+
     let selector = cmd
         .get("selector")
         .and_then(|v| v.as_str())
@@ -3547,6 +3671,32 @@ async fn handle_gettext(cmd: &Value, state: &mut DaemonState) -> Result<Value, S
     Ok(json!({ "text": text, "origin": url }))
 }
 
+async fn handle_frames(_cmd: &Value, state: &mut DaemonState) -> Result<Value, String> {
+    let mgr = state.browser.as_ref().ok_or("Browser not launched")?;
+    let session_id = mgr.active_session_id()?.to_string();
+    let frames = super::element::collect_all_frames_text(
+        &mgr.client,
+        &session_id,
+        &state.iframe_sessions,
+    )
+    .await?;
+    let list: Vec<Value> = frames
+        .iter()
+        .enumerate()
+        .map(|(i, f)| {
+            json!({
+                "index": i,
+                "kind": f.kind,
+                "url": f.url,
+                "frameId": f.frame_id,
+                "textLen": f.text.trim().chars().count(),
+            })
+        })
+        .collect();
+    let url = mgr.get_url().await.unwrap_or_default();
+    Ok(json!({ "frames": list, "count": list.len(), "origin": url }))
+}
+
 async fn handle_getattribute(cmd: &Value, state: &mut DaemonState) -> Result<Value, String> {
     let mgr = state.browser.as_ref().ok_or("Browser not launched")?;
     let session_id = mgr.active_session_id()?.to_string();
@@ -3981,6 +4131,108 @@ async fn handle_cookies_clear(state: &DaemonState) -> Result<Value, String> {
     Ok(json!({ "cleared": true }))
 }
 
+// Detect whether the active page is *currently* a Cloudflare challenge
+// (the full-page "Just a moment…" / "正在进行安全验证" interstitial), so an agent
+// knows whether it must solve or can proceed. Runs in the top frame main world.
+const CF_CHALLENGE_JS: &str = r#"(function(){
+  var t = document.title || '';
+  var challenged =
+    /just a moment|attention required|checking (your|if)|verify you are human|正在进行安全验证|安全验证|请稍候|请完成|人机验证/i.test(t) ||
+    !!document.querySelector('#challenge-form, #challenge-running, #cf-challenge-running, [id^="cf-chl"], script[src*="/cdn-cgi/challenge-platform/"]');
+  var turnstile = !!document.querySelector('.cf-turnstile, [data-sitekey]');
+  return JSON.stringify({ title: t, challenged: challenged, turnstile: turnstile, readyState: document.readyState });
+})()"#;
+
+/// Recommendation for a Cloudflare-gated page, from the current challenge state
+/// and whether a still-valid `cf_clearance` exists. Pure so it's unit-testable.
+/// - not challenged            → "proceed" (the page is cleared/loaded)
+/// - challenged, valid cookie  → "reissue" (clearance present but page still
+///   blocks → it's stale or the IP/UA no longer matches what it was issued for)
+/// - challenged, no cookie     → "solve"
+fn cf_recommendation(challenged: bool, clearance_valid: bool) -> &'static str {
+    if !challenged {
+        "proceed"
+    } else if clearance_valid {
+        "reissue"
+    } else {
+        "solve"
+    }
+}
+
+/// `cf_clearance` validity for a cookie's expiry (epoch seconds; <=0 = session
+/// cookie, treated as non-expiring). Returns (present, expired). Pure.
+fn clearance_state(expires: Option<f64>, now: f64) -> (bool, bool) {
+    match expires {
+        None => (false, false),
+        Some(e) if e <= 0.0 => (true, false), // session cookie: no expiry
+        Some(e) => (true, e < now),
+    }
+}
+
+async fn handle_cf_status(_cmd: &Value, state: &mut DaemonState) -> Result<Value, String> {
+    let mgr = state.browser.as_ref().ok_or("Browser not launched")?;
+    let session_id = mgr.active_session_id()?.to_string();
+    let url = mgr.get_url().await.unwrap_or_default();
+
+    // 1. Is the page a Cloudflare challenge right now?
+    let probe_raw = mgr.evaluate(CF_CHALLENGE_JS, None).await.unwrap_or(Value::Null);
+    let probe = parse_json_string(probe_raw, "cf challenge probe").unwrap_or(Value::Null);
+    let challenged = probe
+        .get("challenged")
+        .and_then(|v| v.as_bool())
+        .unwrap_or(false);
+    let turnstile = probe
+        .get("turnstile")
+        .and_then(|v| v.as_bool())
+        .unwrap_or(false);
+    let title = probe
+        .get("title")
+        .and_then(|v| v.as_str())
+        .unwrap_or("")
+        .to_string();
+
+    // 2. Persistence artifacts: cf_clearance (HttpOnly → must read via CDP, not
+    //    document.cookie) + CF_VERIFIED_DEVICE. Scope to the current URL.
+    let urls = if url.is_empty() {
+        None
+    } else {
+        Some(vec![url.clone()])
+    };
+    let cookies = super::cookies::get_cookies(&mgr.client, &session_id, urls)
+        .await
+        .unwrap_or_default();
+    let clearance = cookies.iter().find(|c| c.name == "cf_clearance");
+    let now = std::time::SystemTime::now()
+        .duration_since(std::time::UNIX_EPOCH)
+        .map(|d| d.as_secs_f64())
+        .unwrap_or(0.0);
+    let (present, expired) = clearance_state(clearance.map(|c| c.expires), now);
+    let expires_in = clearance
+        .filter(|_| present && !expired)
+        .map(|c| (c.expires - now).max(0.0) as i64);
+    let device_verified = cookies
+        .iter()
+        .any(|c| c.name.starts_with("CF_VERIFIED_DEVICE"));
+
+    let clearance_valid = present && !expired;
+    let recommendation = cf_recommendation(challenged, clearance_valid);
+
+    Ok(json!({
+        "url": url,
+        "title": title,
+        "challenged": challenged,
+        "turnstile": turnstile,
+        "clearance": {
+            "present": present,
+            "expired": expired,
+            "expiresIn": expires_in,
+            "httpOnly": clearance.map(|c| c.http_only).unwrap_or(false),
+        },
+        "deviceVerified": device_verified,
+        "recommendation": recommendation,
+    }))
+}
+
 async fn handle_storage_get(cmd: &Value, state: &DaemonState) -> Result<Value, String> {
     let mgr = state.browser.as_ref().ok_or("Browser not launched")?;
     let session_id = mgr.active_session_id()?.to_string();
@@ -4429,7 +4681,33 @@ async fn handle_tab_switch(cmd: &Value, state: &mut DaemonState) -> Result<Value
     state.ref_map.clear();
     state.iframe_sessions.clear();
     state.active_frame_id = None;
-    let result = mgr.tab_switch_by_id(tab_id).await?;
+    let mut result = mgr.tab_switch_by_id(tab_id).await?;
+
+    // Liveness probe: confirm the new session actually answers before we report
+    // success, so `tab <id>` doesn't print a misleading ✓ for a session that's
+    // stale and will fail on the very next command (issue #29.3). On the churned
+    // -tabId case the ext-0.4.9 targetId recovery (#24) self-heals within ~6s, so
+    // we surface a warning rather than a hard error to avoid a false failure
+    // during that window.
+    if mgr.evaluate("1", None).await.is_err() {
+        if let Some(obj) = result.as_object_mut() {
+            obj.insert(
+                "warning".to_string(),
+                json!("switched tab is not responding yet (session re-attaching); retry the next command"),
+            );
+        }
+    }
+
+    // `--activate`: raise this tab to the foreground (the switch made it active;
+    // bring_to_front acts on the active tab) — for handing a specific tab to the
+    // human (issue #24-C). Best-effort; don't fail the switch if it can't.
+    if cmd
+        .get("activate")
+        .and_then(|v| v.as_bool())
+        .unwrap_or(false)
+    {
+        let _ = mgr.bring_to_front().await;
+    }
 
     if let Some(ref server) = state.stream_server {
         if let Ok(dims) = mgr
@@ -5259,6 +5537,18 @@ async fn handle_bringtofront(state: &DaemonState) -> Result<Value, String> {
     Ok(json!({ "broughtToFront": true }))
 }
 
+async fn handle_current(state: &mut DaemonState) -> Result<Value, String> {
+    let mgr = state.browser.as_mut().ok_or("Browser not launched")?;
+    // Refresh so `current` reflects the live URL/title even after a cross-process
+    // nav (the relay's cached target_info can lag) (#26).
+    mgr.resync_targets().await.ok();
+    let mut info = mgr.active_page_info().ok_or("No active tab")?;
+    if let Some(obj) = info.as_object_mut() {
+        obj.insert("current".to_string(), json!(true));
+    }
+    Ok(info)
+}
+
 async fn handle_timezone(cmd: &Value, state: &DaemonState) -> Result<Value, String> {
     let mgr = state.browser.as_ref().ok_or("Browser not launched")?;
     let timezone = cmd
@@ -7164,6 +7454,7 @@ async fn handle_diff_screenshot(cmd: &Value, state: &DaemonState) -> Result<Valu
         quality: None,
         annotate: false,
         output_dir: None,
+        clip: None,
     };
 
     let result = screenshot::take_screenshot(
@@ -8760,13 +9051,7 @@ async fn handle_keydown(cmd: &Value, state: &DaemonState) -> Result<Value, Strin
         .and_then(|v| v.as_str())
         .ok_or("Missing 'key' parameter")?;
 
-    mgr.client
-        .send_command(
-            "Input.dispatchKeyEvent",
-            Some(json!({ "type": "keyDown", "key": key })),
-            Some(&session_id),
-        )
-        .await?;
+    interaction::dispatch_single_key(&mgr.client, &session_id, key, "keyDown").await?;
     Ok(json!({ "keydown": key }))
 }
 
@@ -8778,13 +9063,7 @@ async fn handle_keyup(cmd: &Value, state: &DaemonState) -> Result<Value, String>
         .and_then(|v| v.as_str())
         .ok_or("Missing 'key' parameter")?;
 
-    mgr.client
-        .send_command(
-            "Input.dispatchKeyEvent",
-            Some(json!({ "type": "keyUp", "key": key })),
-            Some(&session_id),
-        )
-        .await?;
+    interaction::dispatch_single_key(&mgr.client, &session_id, key, "keyUp").await?;
     Ok(json!({ "keyup": key }))
 }
 
@@ -8902,6 +9181,26 @@ mod tests {
     use crate::test_utils::EnvGuard;
     use std::fs;
 
+    #[test]
+    fn test_cf_recommendation() {
+        assert_eq!(cf_recommendation(false, false), "proceed");
+        assert_eq!(cf_recommendation(false, true), "proceed");
+        assert_eq!(cf_recommendation(true, false), "solve");
+        assert_eq!(cf_recommendation(true, true), "reissue");
+    }
+
+    #[test]
+    fn test_clearance_state() {
+        // no cookie
+        assert_eq!(clearance_state(None, 1000.0), (false, false));
+        // session cookie (expires <= 0) → present, never expired
+        assert_eq!(clearance_state(Some(-1.0), 1000.0), (true, false));
+        // valid: expiry in the future
+        assert_eq!(clearance_state(Some(2000.0), 1000.0), (true, false));
+        // expired: expiry in the past
+        assert_eq!(clearance_state(Some(500.0), 1000.0), (true, true));
+    }
+
     #[test]
     fn test_url_glob_to_regex() {
         assert_eq!(url_glob_to_regex("**/dashboard"), "^.*/dashboard$");

cli/src/native/browser.rs

@@ -121,7 +121,7 @@ fn normalize_url_for_match(url: &str) -> String {
 fn update_page_target_info_in_pages(pages: &mut [PageInfo], target: &TargetInfo) -> bool {
     if let Some(page) = pages.iter_mut().find(|p| p.target_id == target.target_id) {
         page.url = target.url.clone();
-        page.title = target.title.clone();
+        page.title = sanitize_title(&target.title);
         page.target_type = target.target_type.clone();
         return true;
     }
@@ -166,6 +166,94 @@ fn resolve_active_index(
     active_page_index
 }
 
+/// Strip zero-width / invisible / bidi-format Unicode from a page title before
+/// we store it. Some sites prepend runs of ZWJ / word-joiner / invisible-times /
+/// BOM to `document.title` (badging, watermarking, anti-scrape); left in, they
+/// pollute `tab list`, break text matching, and wreck column alignment (#33).
+fn sanitize_title(s: &str) -> String {
+    s.chars()
+        .filter(|&c| {
+            !matches!(c as u32,
+                0x00AD            // soft hyphen
+                | 0x200B..=0x200F // ZWSP, ZWNJ, ZWJ, LRM, RLM
+                | 0x2028 | 0x2029 // line / paragraph separators
+                | 0x202A..=0x202E // bidi embedding/override
+                | 0x2060..=0x2064 // word joiner, invisible operators
+                | 0x2066..=0x2069 // bidi isolates
+                | 0x180E          // Mongolian vowel separator
+                | 0xFEFF          // BOM / ZW no-break space
+            )
+        })
+        .collect::<String>()
+        .trim()
+        .to_string()
+}
+
+/// Best-effort MIME type from a filename extension, for the relay file-upload
+/// fallback (the page-constructed `File` needs a sensible `type`). Covers the
+/// common upload kinds; anything unknown falls back to a generic binary type.
+fn mime_for_path(name: &str) -> &'static str {
+    let ext = name.rsplit('.').next().unwrap_or("").to_lowercase();
+    match ext.as_str() {
+        "png" => "image/png",
+        "jpg" | "jpeg" => "image/jpeg",
+        "gif" => "image/gif",
+        "webp" => "image/webp",
+        "svg" => "image/svg+xml",
+        "bmp" => "image/bmp",
+        "pdf" => "application/pdf",
+        "txt" => "text/plain",
+        "csv" => "text/csv",
+        "json" => "application/json",
+        "mp4" => "video/mp4",
+        "webm" => "video/webm",
+        "mov" => "video/quicktime",
+        "mp3" => "audio/mpeg",
+        "zip" => "application/zip",
+        _ => "application/octet-stream",
+    }
+}
+
+/// Target ids to prune after a `Target.getTargets` resync: tracked pages whose
+/// target is no longer in the live set — EXCEPT the explicitly-pinned active
+/// target, which is protected. The relay against a busy real Chrome occasionally
+/// returns a different window's tabs for a single `getTargets` call ("tab list
+/// hops windows", issue #31); pruning on that transient snapshot would drop the
+/// agent's adopted tab and drift subsequent eval/click onto a foreign tab. A
+/// genuine close still arrives as `Target.targetDestroyed` (handled in the event
+/// drain), which removes the pin properly — so protecting it here only guards
+/// against flaky snapshots, not real closures.
+fn prunable_target_ids(
+    pages: &[PageInfo],
+    live_ids: &HashSet<String>,
+    pinned: Option<&str>,
+) -> Vec<String> {
+    pages
+        .iter()
+        .map(|p| p.target_id.clone())
+        .filter(|tid| !live_ids.contains(tid) && pinned != Some(tid.as_str()))
+        .collect()
+}
+
+/// Whether the resolved active page is a tab the session created (its target_id
+/// is in `created_targets`). Pure core of [`BrowserManager::active_is_session_owned`]
+/// so the relay no-hijack rule is unit-testable without a live browser.
+fn active_index_is_owned(
+    pages: &[PageInfo],
+    active_target_id: Option<&str>,
+    active_page_index: usize,
+    created_targets: &HashSet<String>,
+) -> bool {
+    pages
+        .get(resolve_active_index(
+            pages,
+            active_target_id,
+            active_page_index,
+        ))
+        .map(|p| created_targets.contains(&p.target_id))
+        .unwrap_or(false)
+}
+
 /// Converts common error messages into AI-friendly, actionable descriptions.
 pub fn to_ai_friendly_error(error: &str) -> String {
     let lower = error.to_lowercase();
@@ -471,6 +559,13 @@ impl BrowserManager {
             }
         };
 
+        // A launched browser carries a debug port → it's the other path that can
+        // pop Chrome's consent modal; record it for #31 diagnosis.
+        crate::connect::log_connect_mode(
+            &ws_url,
+            true,
+            DAEMON_SESSION.get().map(String::as_str).unwrap_or("default"),
+        );
         let manager = if engine == "lightpanda" {
             initialize_lightpanda_manager(ws_url, process).await?
         } else {
@@ -566,6 +661,13 @@ impl BrowserManager {
         headers: Option<Vec<(String, String)>>,
     ) -> Result<Self, String> {
         let ws_url = resolve_cdp_url(url).await?;
+        // Record the transport so a reappearing "Allow remote debugging?" modal
+        // can be traced to a raw-port attach vs the consent-free relay (#31).
+        crate::connect::log_connect_mode(
+            &ws_url,
+            false,
+            DAEMON_SESSION.get().map(String::as_str).unwrap_or("default"),
+        );
         let client = Arc::new(CdpClient::connect_with_headers(&ws_url, headers).await?);
         let mut manager = Self {
             client,
@@ -694,7 +796,7 @@ impl BrowserManager {
                     target_id: target.target_id.clone(),
                     session_id: attach_result.session_id.clone(),
                     url: target.url.clone(),
-                    title: target.title.clone(),
+                    title: sanitize_title(&target.title),
                     target_type: target.target_type.clone(),
                 });
             }
@@ -799,6 +901,20 @@ impl BrowserManager {
         )
     }
 
+    /// Whether the resolved active page is a tab THIS session created (via
+    /// `Target.createTarget` — `tab new`, `ensure_page`, or the first `open`).
+    /// On the shared real browser a fresh session also passively attaches to the
+    /// user's existing tabs; those are NOT owned, and navigating one would
+    /// clobber the user's page. Used to gate `navigate` on the relay.
+    fn active_is_session_owned(&self) -> bool {
+        active_index_is_owned(
+            &self.pages,
+            self.active_target_id.as_deref(),
+            self.active_page_index,
+            &self.created_targets,
+        )
+    }
+
     /// Pin the current active page by target_id so later commands stick to it.
     /// Call after any explicit open / tab new / tab switch.
     fn pin_active_target(&mut self) {
@@ -816,6 +932,18 @@ impl BrowserManager {
     }
 
     pub async fn navigate(&mut self, url: &str, wait_until: WaitUntil) -> Result<Value, String> {
+        // On the shared real browser (extension relay), a fresh session only
+        // passively attached to the user's existing tabs — it doesn't own any. The
+        // pre-fix code made one of those the active tab, so the first `open` then
+        // navigated (clobbered) the user's page: in dogfooding an `open` replaced a
+        // half-filled form with the target site. If the active tab isn't one we
+        // created, open our own tab in this session's group and navigate THAT, so
+        // the user's (and other sessions') tabs are never hijacked. Off the relay
+        // (a browser we launched) reusing the active tab is correct, so this is
+        // gated on `agent_group()`.
+        if self.agent_group().is_some() && !self.active_is_session_owned() {
+            self.tab_new(None, None).await?;
+        }
         let session_id = self.active_session_id()?.to_string();
         let mut lifecycle_rx = self.client.subscribe();
 
@@ -894,7 +1022,7 @@ impl BrowserManager {
         self.active_page_index = self.resolved_active_index();
         if let Some(page) = self.pages.get_mut(self.active_page_index) {
             page.url = page_url.clone();
-            page.title = title.clone();
+            page.title = sanitize_title(&title);
         }
         self.pin_active_target();
 
@@ -956,7 +1084,7 @@ impl BrowserManager {
 
     pub async fn get_title(&self) -> Result<String, String> {
         let result = self.evaluate_simple("document.title").await?;
-        Ok(result.as_str().unwrap_or("").to_string())
+        Ok(sanitize_title(result.as_str().unwrap_or("")))
     }
 
     pub async fn get_content(&self) -> Result<String, String> {
@@ -1218,6 +1346,22 @@ impl BrowserManager {
             .collect()
     }
 
+    /// The active tab's stable handle + current location, for `chrome-use
+    /// current` (#26). `targetId` survives cross-process navigation, so it's the
+    /// handle an agent should hold across a multi-step flow.
+    pub fn active_page_info(&self) -> Option<Value> {
+        let i = self.resolved_active_index();
+        self.pages.get(i).map(|p| {
+            json!({
+                "tabId": format_tab_id(p.tab_id),
+                "targetId": p.target_id,
+                "label": p.label,
+                "url": p.url,
+                "title": p.title,
+            })
+        })
+    }
+
     /// Stable `tab_id` for a page identified by its CDP `targetId`, if tracked.
     /// Lets callers adopt a tab by the cross-session-stable target id.
     pub fn tab_id_for_target(&self, target_id: &str) -> Option<u32> {
@@ -1234,6 +1378,67 @@ impl BrowserManager {
     /// the active tab is preserved, and re-pinned if it was pruned. Powers a live
     /// `tab list` and adopt-by-targetId so a fresh session can reach a stranded,
     /// still-filled tab without reloading it (issue #21).
+    /// Detect targets that appeared since the `before` set (e.g. a click that
+    /// opened a new tab via a `target=_blank` link or `window.open`), attach +
+    /// track each in the background, and return the first newly-opened page.
+    ///
+    /// Lighter than [`resync_targets`] — one `getTargets` and work only on the
+    /// new targets, no whole-tab url/title refresh — so it's cheap enough to run
+    /// after every click. The new tab is added in the background (never steals
+    /// the active tab, per #7/#8.1); the caller surfaces it so the agent knows a
+    /// tab opened instead of seeing the old page (issue #24-A).
+    pub async fn adopt_newly_opened(&mut self, before: &HashSet<String>) -> Option<PageInfo> {
+        let result: GetTargetsResult = self
+            .client
+            .send_command_typed("Target.getTargets", &json!({}), None)
+            .await
+            .ok()?;
+        let live: Vec<TargetInfo> = result
+            .target_infos
+            .into_iter()
+            .filter(should_track_target)
+            .collect();
+        let mut opened: Option<PageInfo> = None;
+        for target in &live {
+            if before.contains(&target.target_id)
+                || self.pages.iter().any(|p| p.target_id == target.target_id)
+            {
+                continue;
+            }
+            let attach: AttachToTargetResult = match self
+                .client
+                .send_command_typed(
+                    "Target.attachToTarget",
+                    &AttachToTargetParams {
+                        target_id: target.target_id.clone(),
+                        flatten: true,
+                    },
+                    None,
+                )
+                .await
+            {
+                Ok(r) => r,
+                Err(_) => continue,
+            };
+            let tab_id = self.assign_tab_id();
+            let page = PageInfo {
+                tab_id,
+                label: None,
+                target_id: target.target_id.clone(),
+                session_id: attach.session_id.clone(),
+                url: target.url.clone(),
+                title: sanitize_title(&target.title),
+                target_type: target.target_type.clone(),
+            };
+            self.add_background_page(page.clone());
+            let _ = self.enable_domains(&attach.session_id).await;
+            if opened.is_none() {
+                opened = Some(page);
+            }
+        }
+        opened
+    }
+
     pub async fn resync_targets(&mut self) -> Result<(), String> {
         self.client
             .send_command_typed::<_, Value>(
@@ -1283,19 +1488,16 @@ impl BrowserManager {
                 target_id: target.target_id.clone(),
                 session_id: attach_result.session_id.clone(),
                 url: target.url.clone(),
-                title: target.title.clone(),
+                title: sanitize_title(&target.title),
                 target_type: target.target_type.clone(),
             });
             let _ = self.enable_domains(&attach_result.session_id).await;
         }
 
-        // Drop tabs that no longer exist so `tab list` doesn't show phantom rows.
-        let gone: Vec<String> = self
-            .pages
-            .iter()
-            .map(|p| p.target_id.clone())
-            .filter(|tid| !live_ids.contains(tid))
-            .collect();
+        // Drop tabs that no longer exist so `tab list` doesn't show phantom rows —
+        // but never prune the explicitly-pinned active target on a transient
+        // getTargets snapshot (issue #31; see `prunable_target_ids`).
+        let gone = prunable_target_ids(&self.pages, &live_ids, self.active_target_id.as_deref());
         for tid in gone {
             self.remove_page_by_target_id(&tid);
         }
@@ -1329,7 +1531,7 @@ impl BrowserManager {
                             }
                         }
                         if let Some(t) = ti.get("title").and_then(|v| v.as_str()) {
-                            page.title = t.to_string();
+                            page.title = sanitize_title(t);
                         }
                     }
                 }
@@ -1517,7 +1719,7 @@ impl BrowserManager {
 
         if let Some(page) = self.pages.get_mut(index) {
             page.url = url.clone();
-            page.title = title.clone();
+            page.title = sanitize_title(&title);
         }
 
         let page = &self.pages[index];
@@ -1772,7 +1974,8 @@ impl BrowserManager {
             .and_then(|v| v.as_i64())
             .ok_or("Could not get backendNodeId for file input")?;
 
-        self.client
+        let set_files = self
+            .client
             .send_command(
                 "DOM.setFileInputFiles",
                 Some(json!({
@@ -1781,26 +1984,153 @@ impl BrowserManager {
                 })),
                 Some(&effective_session_id),
             )
-            .await
-            .map_err(|e| {
-                // Chrome's chrome.debugger API (the extension-relay transport)
-                // forbids DOM.setFileInputFiles for security, surfacing as an
-                // opaque `-32000 "Not allowed"`. Translate it into an actionable
-                // message rather than leaking the raw CDP error (issue #13).
-                if e.contains("Not allowed") || e.contains("-32000") {
-                    "file upload isn't supported over the extension relay — \
-                     Chrome's chrome.debugger API forbids DOM.setFileInputFiles. \
-                     Use a direct-CDP session instead: \
-                     `chrome-use --session up --launch open <url>` (carry your \
-                     login over with `cookies export` | `cookies set --curl`), \
-                     then run `upload` in that session. \
-                     See https://github.com/leeguooooo/chrome-use/issues/13"
-                        .to_string()
-                } else {
-                    e
-                }
-            })?;
+            .await;
 
+        if let Err(e) = set_files {
+            // Chrome's chrome.debugger API (the extension-relay transport) forbids
+            // DOM.setFileInputFiles for security, surfacing as an opaque
+            // `-32000 "Not allowed"`. Fall back to constructing the File entirely
+            // IN THE PAGE and assigning it to the input — the standard
+            // Playwright/Cypress trick, which needs no privileged CDP and so works
+            // over the relay (issue #13).
+            if e.contains("Not allowed") || e.contains("-32000") {
+                return self
+                    .upload_files_via_page(object_id, files, &effective_session_id)
+                    .await;
+            }
+            return Err(e);
+        }
+
+        Ok(())
+    }
+
+    /// Relay-safe file upload: read each file locally, hand its bytes to the page
+    /// as base64, and rebuild a `File` there — then either assign it to a file
+    /// `<input>` (Chrome allows `input.files = dataTransfer.files`) or, for a
+    /// dropzone/composer, dispatch synthetic `paste`/`drop` events carrying the
+    /// `DataTransfer`. No `DOM.setFileInputFiles`, so chrome.debugger permits it.
+    async fn upload_files_via_page(
+        &self,
+        object_id: String,
+        files: &[String],
+        session_id: &str,
+    ) -> Result<(), String> {
+        use base64::Engine;
+        // The relay tunnels every CDP message through Chrome native messaging,
+        // which caps a single message at ~1 MiB. A whole image's base64 blows
+        // past that ("CDP response channel closed"), so we STREAM the bytes into
+        // a page-side buffer in sub-limit chunks, then assemble the File from it.
+        const CHUNK: usize = 96 * 1024; // base64 chars per message; safe under 1 MiB
+
+        // Reset the staging buffer.
+        self.client
+            .send_command(
+                "Runtime.evaluate",
+                Some(json!({ "expression": "window.__cuUpload = [];", "returnByValue": true })),
+                Some(session_id),
+            )
+            .await
+            .map_err(|e| format!("relay upload (reset) failed: {}", e))?;
+
+        for path in files {
+            let bytes = std::fs::read(path).map_err(|e| format!("cannot read {}: {}", path, e))?;
+            let name = std::path::Path::new(path)
+                .file_name()
+                .and_then(|n| n.to_str())
+                .unwrap_or("upload.bin")
+                .to_string();
+            let mime = mime_for_path(&name);
+            let b64 = base64::engine::general_purpose::STANDARD.encode(&bytes);
+
+            // Push the file's metadata with an empty buffer.
+            let init = format!(
+                "window.__cuUpload.push({{ name: {}, type: {}, b64: '' }});",
+                serde_json::to_string(&name).unwrap_or_default(),
+                serde_json::to_string(mime).unwrap_or_default(),
+            );
+            self.client
+                .send_command(
+                    "Runtime.evaluate",
+                    Some(json!({ "expression": init, "returnByValue": true })),
+                    Some(session_id),
+                )
+                .await
+                .map_err(|e| format!("relay upload (init) failed: {}", e))?;
+
+            // Stream the base64 in chunks. base64's alphabet (A–Za–z0–9+/=) needs
+            // no escaping inside a single-quoted JS string, so concatenation is safe.
+            let idx = "window.__cuUpload[window.__cuUpload.length-1].b64";
+            let mut start = 0;
+            while start < b64.len() {
+                let end = (start + CHUNK).min(b64.len());
+                let chunk = &b64[start..end];
+                let expr = format!("{idx} += '{chunk}';");
+                self.client
+                    .send_command(
+                        "Runtime.evaluate",
+                        Some(json!({ "expression": expr, "returnByValue": true })),
+                        Some(session_id),
+                    )
+                    .await
+                    .map_err(|e| format!("relay upload (chunk) failed: {}", e))?;
+                start = end;
+            }
+        }
+
+        // Assemble the Files from the buffer and attach to the element, then clean up.
+        let func = r#"function() {
+            const filesData = window.__cuUpload || [];
+            const dt = new DataTransfer();
+            for (const f of filesData) {
+                const bin = atob(f.b64);
+                const arr = new Uint8Array(bin.length);
+                for (let i = 0; i < bin.length; i++) arr[i] = bin.charCodeAt(i);
+                dt.items.add(new File([arr], f.name, { type: f.type }));
+            }
+            try { delete window.__cuUpload; } catch (e) { window.__cuUpload = undefined; }
+            const el = this;
+            if (el.tagName === 'INPUT' && el.type === 'file') {
+                el.files = dt.files;
+                el.dispatchEvent(new Event('input', { bubbles: true }));
+                el.dispatchEvent(new Event('change', { bubbles: true }));
+                return 'input:' + dt.files.length;
+            }
+            // Dropzone / rich composer: replay paste then drop with the files.
+            try { el.dispatchEvent(new ClipboardEvent('paste', { bubbles: true, clipboardData: dt })); } catch (e) {}
+            try {
+                const ev = new DragEvent('drop', { bubbles: true, cancelable: true });
+                Object.defineProperty(ev, 'dataTransfer', { value: dt });
+                el.dispatchEvent(ev);
+            } catch (e) {}
+            return 'event:' + dt.files.length;
+        }"#;
+
+        let result: EvaluateResult = self
+            .client
+            .send_command_typed(
+                "Runtime.callFunctionOn",
+                &CallFunctionOnParams {
+                    function_declaration: func.to_string(),
+                    object_id: Some(object_id),
+                    arguments: None,
+                    return_by_value: Some(true),
+                    await_promise: Some(false),
+                },
+                Some(session_id),
+            )
+            .await
+            .map_err(|e| format!("relay file-injection failed: {}", e))?;
+
+        if let Some(ref details) = result.exception_details {
+            return Err(format!(
+                "relay file-injection threw: {}",
+                details
+                    .exception
+                    .as_ref()
+                    .and_then(|ex| ex.description.as_deref())
+                    .unwrap_or(&details.text)
+            ));
+        }
         Ok(())
     }
 
@@ -2431,6 +2761,77 @@ mod tests {
         assert_eq!(resolve_active_index(&pages, Some("CLOSED"), 1), 1);
     }
 
+    // --- issue: `open` must not hijack a user's tab on the relay (dogfood) ---
+
+    #[test]
+    fn active_not_owned_when_only_user_tabs_discovered() {
+        // A fresh relay session passively attached to the user's tabs but created
+        // none — so navigate must NOT reuse the active tab (it'd clobber the
+        // user's page); it has to open its own first.
+        let pages = vec![page("USER_A"), page("USER_B")];
+        let created = HashSet::new();
+        assert!(!active_index_is_owned(&pages, Some("USER_A"), 0, &created));
+    }
+
+    #[test]
+    fn active_owned_when_session_created_the_tab() {
+        let pages = vec![page("USER_A"), page("OURS")];
+        let mut created = HashSet::new();
+        created.insert("OURS".to_string());
+        // Active pinned to the tab we created → safe to navigate it.
+        assert!(active_index_is_owned(&pages, Some("OURS"), 1, &created));
+        // But pinned to the user's tab → not owned, even though we own another.
+        assert!(!active_index_is_owned(&pages, Some("USER_A"), 0, &created));
+    }
+
+    #[test]
+    fn active_not_owned_when_no_pages() {
+        let created = HashSet::new();
+        assert!(!active_index_is_owned(&[], None, 0, &created));
+    }
+
+    #[test]
+    fn test_sanitize_title() {
+        // The exact pollution from #33: ZWJ / word-joiner / invisible-times / BOM
+        // prepended to "GitHub".
+        let dirty = "\u{200d}\u{2061}\u{200d}\u{2063}\u{200b}\u{2062}\u{feff}GitHub";
+        assert_eq!(sanitize_title(dirty), "GitHub");
+        // Clean titles (incl. CJK + normal punctuation) pass through untouched.
+        assert_eq!(sanitize_title("購入手続きへ - メルカリ"), "購入手続きへ - メルカリ");
+        assert_eq!(sanitize_title("  Hello World  "), "Hello World");
+        // Emoji and real content survive; only the invisibles are dropped.
+        assert_eq!(sanitize_title("✓ Done\u{200b}"), "✓ Done");
+    }
+
+    #[test]
+    fn test_mime_for_path() {
+        assert_eq!(mime_for_path("a.png"), "image/png");
+        assert_eq!(mime_for_path("PHOTO.JPG"), "image/jpeg");
+        assert_eq!(mime_for_path("clip.webp"), "image/webp");
+        assert_eq!(mime_for_path("doc.pdf"), "application/pdf");
+        assert_eq!(mime_for_path("noext"), "application/octet-stream");
+        assert_eq!(mime_for_path("weird.xyz"), "application/octet-stream");
+    }
+
+    #[test]
+    fn prune_protects_pinned_target_on_transient_snapshot() {
+        // The relay returned a getTargets snapshot missing the pinned tab "A"
+        // (it hopped to another window). "B" is also absent. Without protection
+        // both would be pruned and the next command would drift; with the pin
+        // protected, only the genuinely-unpinned "B" is dropped (issue #31).
+        let pages = vec![page("A"), page("B")];
+        let live: HashSet<String> = HashSet::new(); // snapshot returned neither
+        let gone = prunable_target_ids(&pages, &live, Some("A"));
+        assert_eq!(gone, vec!["B".to_string()]);
+        // With no pin, both are prunable (unchanged behavior).
+        let gone_unpinned = prunable_target_ids(&pages, &live, None);
+        assert_eq!(gone_unpinned.len(), 2);
+        // A pinned target that IS in the live set is simply not prunable anyway.
+        let mut live2 = HashSet::new();
+        live2.insert("A".to_string());
+        assert_eq!(prunable_target_ids(&pages, &live2, Some("A")), vec!["B".to_string()]);
+    }
+
     #[test]
     fn resolve_active_index_pin_survives_passive_background_tab() {
         // A foreign tab ("Z") gets appended by passive discovery after we pinned

cli/src/native/element.rs

@@ -796,16 +796,43 @@ pub(super) fn extract_ax_string(value: &Option<AXValue>) -> String {
 /// Build a JS expression that finds a DOM element by CSS selector or XPath.
 fn build_find_element_js(selector: &str) -> String {
     if let Some(xpath) = selector.strip_prefix("xpath=") {
-        format!(
+        return format!(
             "document.evaluate({}, document, null, XPathResult.FIRST_ORDERED_NODE_TYPE, null).singleNodeValue",
             serde_json::to_string(xpath).unwrap_or_default()
-        )
-    } else {
-        format!(
-            "document.querySelector({})",
-            serde_json::to_string(selector).unwrap_or_default()
-        )
+        );
     }
+    // Bare string (or explicit `text=`): try CSS first, then fall back to
+    // matching an interactive element by its VISIBLE TEXT. snapshot exposes
+    // buttons/links by their name, so `click "購入手続きへ"` should resolve by
+    // that label — previously it was fed straight to `querySelector` as CSS and
+    // failed as an invalid selector even though the button was right there
+    // (issue #24-B). CSS still wins when it matches, so existing selectors are
+    // unaffected; nested/non-ASCII labels now resolve too.
+    let text_only = selector.strip_prefix("text=");
+    let force_text = text_only.is_some();
+    let sel_json = serde_json::to_string(selector).unwrap_or_default();
+    let want_json = serde_json::to_string(text_only.unwrap_or(selector)).unwrap_or_default();
+    format!(
+        r#"(() => {{
+  const sel = {sel};
+  const css = {force_text} ? null : (() => {{ try {{ return document.querySelector(sel); }} catch (_e) {{ return null; }} }})();
+  if (css) return css;
+  const norm = s => (s == null ? '' : String(s)).replace(/\s+/g, ' ').trim();
+  const w = norm({want}); if (!w) return null;
+  const wl = w.toLowerCase();
+  const interactive = Array.from(document.querySelectorAll(
+    'button,a,[role=button],[role=link],[role=menuitem],[role=tab],[role=option],input[type=submit],input[type=button],input[type=reset],summary,label,[onclick]'));
+  const textOf = e => norm(e.innerText || e.textContent) || norm(e.value) ||
+    norm(e.getAttribute && e.getAttribute('aria-label')) || norm(e.getAttribute && e.getAttribute('title'));
+  let hit = interactive.find(e => textOf(e) === w) || interactive.find(e => textOf(e).toLowerCase().includes(wl));
+  if (hit) return hit;
+  const leaves = Array.from(document.querySelectorAll('*')).filter(e => !e.children.length);
+  return leaves.find(e => norm(e.textContent) === w) || leaves.find(e => norm(e.textContent).toLowerCase().includes(wl)) || null;
+}})()"#,
+        sel = sel_json,
+        want = want_json,
+        force_text = force_text
+    )
 }
 
 /// Build a JS expression that counts matching DOM elements by CSS selector or XPath.
@@ -948,6 +975,263 @@ pub async fn get_element_text(
         .unwrap_or_default())
 }
 
+/// Text content collected from a single frame of the page.
+#[derive(Debug, Clone)]
+pub struct FrameText {
+    pub frame_id: String,
+    pub url: String,
+    /// "top" | "inline" (same-process child frame) | "oopif" (out-of-process).
+    pub kind: &'static str,
+    pub text: String,
+}
+
+// The expression we run in every frame to read its visible text. innerText
+// honors CSS visibility (skips display:none), textContent is the fallback.
+const FRAME_INNERTEXT_JS: &str = "(function(){try{var b=document.body||document.documentElement;return b?(b.innerText||b.textContent||''):'';}catch(e){return '';}})()";
+
+async fn eval_text_default(client: &CdpClient, session_id: &str) -> String {
+    let res = client
+        .send_command(
+            "Runtime.evaluate",
+            Some(serde_json::json!({
+                "expression": FRAME_INNERTEXT_JS,
+                "returnByValue": true,
+            })),
+            Some(session_id),
+        )
+        .await;
+    res.ok()
+        .and_then(|v| v.get("result").and_then(|r| r.get("value")).cloned())
+        .and_then(|v| v.as_str().map(|s| s.to_string()))
+        .unwrap_or_default()
+}
+
+// Same-process child frames share the top renderer but live in their own
+// execution context. Page.createIsolatedWorld hands us a context id bound to
+// that frame so Runtime.evaluate reads the child document, not the parent.
+async fn eval_text_in_frame(client: &CdpClient, session_id: &str, frame_id: &str) -> String {
+    let ctx = client
+        .send_command(
+            "Page.createIsolatedWorld",
+            Some(serde_json::json!({ "frameId": frame_id, "worldName": "chrome_use_text" })),
+            Some(session_id),
+        )
+        .await
+        .ok()
+        .and_then(|v| v.get("executionContextId").and_then(|c| c.as_i64()));
+    let Some(ctx_id) = ctx else { return String::new() };
+    let res = client
+        .send_command(
+            "Runtime.evaluate",
+            Some(serde_json::json!({
+                "expression": FRAME_INNERTEXT_JS,
+                "returnByValue": true,
+                "contextId": ctx_id,
+            })),
+            Some(session_id),
+        )
+        .await;
+    res.ok()
+        .and_then(|v| v.get("result").and_then(|r| r.get("value")).cloned())
+        .and_then(|v| v.as_str().map(|s| s.to_string()))
+        .unwrap_or_default()
+}
+
+fn flatten_frame_tree(node: &Value, is_top: bool, out: &mut Vec<(String, String, bool)>) {
+    if let Some(frame) = node.get("frame") {
+        if let Some(id) = frame.get("id").and_then(|v| v.as_str()) {
+            let url = frame
+                .get("url")
+                .and_then(|v| v.as_str())
+                .unwrap_or("")
+                .to_string();
+            out.push((id.to_string(), url, is_top));
+        }
+    }
+    if let Some(children) = node.get("childFrames").and_then(|v| v.as_array()) {
+        for child in children {
+            flatten_frame_tree(child, false, out);
+        }
+    }
+}
+
+/// Collect visible text from every frame reachable in the active session,
+/// including out-of-process iframes (which never appear in the top frame's
+/// `Page.getFrameTree` and so are invisible to `document.body.innerText`).
+///
+/// Same-process child frames are read through `Page.createIsolatedWorld`;
+/// OOPIFs are read through their own auto-attached debugger session
+/// (`iframe_sessions`, keyed by frameId == targetId). This is the engine
+/// behind `get text --all-frames` and `chrome-use frames` — the fix for
+/// listing/marketplace pages whose description lives in a child frame (#27).
+pub async fn collect_all_frames_text(
+    client: &CdpClient,
+    top_session: &str,
+    iframe_sessions: &HashMap<String, String>,
+) -> Result<Vec<FrameText>, String> {
+    let mut out: Vec<FrameText> = Vec::new();
+    let mut seen: std::collections::HashSet<String> = std::collections::HashSet::new();
+
+    // 1. Top session: the top frame plus its same-process descendants. OOPIF
+    //    frames that happen to surface here are skipped — they're read via
+    //    their dedicated session in step 2 (cross-process isolated worlds fail).
+    let tree = client
+        .send_command_no_params("Page.getFrameTree", Some(top_session))
+        .await?;
+    let mut frames: Vec<(String, String, bool)> = Vec::new();
+    flatten_frame_tree(&tree["frameTree"], true, &mut frames);
+    for (fid, url, is_top) in frames {
+        if iframe_sessions.contains_key(&fid) {
+            continue;
+        }
+        if !seen.insert(fid.clone()) {
+            continue;
+        }
+        let (kind, text) = if is_top {
+            ("top", eval_text_default(client, top_session).await)
+        } else {
+            ("inline", eval_text_in_frame(client, top_session, &fid).await)
+        };
+        out.push(FrameText {
+            frame_id: fid,
+            url,
+            kind,
+            text,
+        });
+    }
+
+    // 2. Each out-of-process iframe, read through its own session.
+    for (fid, sid) in iframe_sessions {
+        if !seen.insert(fid.clone()) {
+            continue;
+        }
+        let url = client
+            .send_command_no_params("Page.getFrameTree", Some(sid))
+            .await
+            .ok()
+            .and_then(|t| {
+                t.get("frameTree")
+                    .and_then(|ft| ft.get("frame"))
+                    .and_then(|f| f.get("url"))
+                    .and_then(|v| v.as_str())
+                    .map(|s| s.to_string())
+            })
+            .unwrap_or_default();
+        let text = eval_text_default(client, sid).await;
+        out.push(FrameText {
+            frame_id: fid.clone(),
+            url,
+            kind: "oopif",
+            text,
+        });
+    }
+
+    Ok(out)
+}
+
+// Readability-lite: prefer the page's semantic main-content region over the
+// whole body so global header/nav/footer chrome (and, on many listing pages,
+// the "related items" sidebar) doesn't drown out the actual content. Runs on
+// the live, rendered tree (innerText needs layout — a detached clone returns
+// empty), so we pick the densest <main>/<article> region rather than cloning
+// and stripping. Falls back to <body> when no substantial main region exists.
+const MAIN_CONTENT_JS: &str = r#"(function(){
+  function txt(el){try{return (el.innerText||'').trim();}catch(e){return '';}}
+  var sels=['main','[role=main]','article','#main','#contents','#l-content'];
+  var best=null,bestLen=0;
+  for(var i=0;i<sels.length;i++){
+    var els=document.querySelectorAll(sels[i]);
+    for(var j=0;j<els.length;j++){var l=txt(els[j]).length;if(l>bestLen){bestLen=l;best=els[j];}}
+  }
+  if(best&&bestLen>200)return txt(best);
+  return txt(document.body);
+})()"#;
+
+/// Extract the page's main-content text (readability-lite), preferring a
+/// semantic `<main>`/`<article>` region over the full body. Used by
+/// `get text --main` to avoid header/nav/sidebar boilerplate (#27).
+pub async fn get_main_content_text(client: &CdpClient, session_id: &str) -> Result<String, String> {
+    let res = client
+        .send_command(
+            "Runtime.evaluate",
+            Some(serde_json::json!({
+                "expression": MAIN_CONTENT_JS,
+                "returnByValue": true,
+            })),
+            Some(session_id),
+        )
+        .await?;
+    Ok(res
+        .get("result")
+        .and_then(|r| r.get("value"))
+        .and_then(|v| v.as_str())
+        .unwrap_or_default()
+        .to_string())
+}
+
+// Text nodes whose parent is one of these carry no visible content.
+fn is_noise_tag(name: &str) -> bool {
+    matches!(name, "SCRIPT" | "STYLE" | "NOSCRIPT" | "TEMPLATE" | "HEAD")
+}
+
+// Walk a CDP DOM.Node tree, collecting text-node values. Unlike `innerText`
+// (JS, blocked by CLOSED shadow roots), the CDP DOM tree from
+// `DOM.getDocument(pierce:true)` includes closed shadow roots and child
+// documents — so this reaches text JS can't. `parent_noise` carries whether an
+// ancestor was <script>/<style>/etc so their text is skipped.
+fn collect_dom_text(node: &Value, parent_noise: bool, out: &mut String) {
+    let node_type = node.get("nodeType").and_then(|v| v.as_i64()).unwrap_or(0);
+    let node_name = node.get("nodeName").and_then(|v| v.as_str()).unwrap_or("");
+    if node_type == 3 {
+        if !parent_noise {
+            if let Some(t) = node.get("nodeValue").and_then(|v| v.as_str()) {
+                let t = t.trim();
+                if !t.is_empty() {
+                    if !out.is_empty() {
+                        out.push(' ');
+                    }
+                    out.push_str(t);
+                }
+            }
+        }
+        return;
+    }
+    let noise = parent_noise || is_noise_tag(node_name);
+    if let Some(children) = node.get("children").and_then(|v| v.as_array()) {
+        for child in children {
+            collect_dom_text(child, noise, out);
+        }
+    }
+    if let Some(shadow) = node.get("shadowRoots").and_then(|v| v.as_array()) {
+        for sr in shadow {
+            collect_dom_text(sr, noise, out);
+        }
+    }
+    if let Some(doc) = node.get("contentDocument") {
+        collect_dom_text(doc, noise, out);
+    }
+}
+
+/// Extract text from the page via the CDP DOM tree with `pierce:true`, which
+/// reaches into CLOSED shadow roots and child documents that `innerText`/`eval`
+/// cannot. Lets an agent read content rendered into a closed shadow DOM (e.g. an
+/// extension's injected debug panel) without any extra Chrome permission — it
+/// rides the per-tab debugger session that's already attached (#30).
+pub async fn get_pierced_text(client: &CdpClient, session_id: &str) -> Result<String, String> {
+    let doc = client
+        .send_command(
+            "DOM.getDocument",
+            Some(serde_json::json!({ "depth": -1, "pierce": true })),
+            Some(session_id),
+        )
+        .await?;
+    let mut out = String::new();
+    if let Some(root) = doc.get("root") {
+        collect_dom_text(root, false, &mut out);
+    }
+    Ok(out)
+}
+
 pub async fn get_element_attribute(
     client: &CdpClient,
     session_id: &str,
@@ -1421,6 +1705,31 @@ mod tests {
         assert_eq!(parse_ref("@e123"), Some("e123".to_string()));
     }
 
+    #[test]
+    fn test_collect_dom_text_pierces_closed_shadow_and_skips_noise() {
+        // A CDP DOM.Node tree: a host element whose CLOSED shadow root holds the
+        // text, plus a <script> whose text must be skipped.
+        let tree = serde_json::json!({
+            "nodeType": 1, "nodeName": "BODY",
+            "children": [
+                { "nodeType": 1, "nodeName": "SCRIPT",
+                  "children": [ { "nodeType": 3, "nodeName": "#text", "nodeValue": "var secret=1;" } ] },
+                { "nodeType": 1, "nodeName": "DIV",
+                  "shadowRoots": [
+                    { "nodeType": 11, "nodeName": "#document-fragment",
+                      "children": [
+                        { "nodeType": 1, "nodeName": "SPAN",
+                          "children": [ { "nodeType": 3, "nodeName": "#text", "nodeValue": "DECRYPTED 42" } ] }
+                      ] }
+                  ] }
+            ]
+        });
+        let mut out = String::new();
+        collect_dom_text(&tree, false, &mut out);
+        assert_eq!(out, "DECRYPTED 42");
+        assert!(!out.contains("secret"), "script text must be skipped");
+    }
+
     #[test]
     fn test_parse_ref_equals_prefix() {
         assert_eq!(parse_ref("ref=e1"), Some("e1".to_string()));
@@ -1452,10 +1761,30 @@ mod tests {
     #[test]
     fn test_build_selector_js_css() {
         let js = build_selector_js("#submit-btn");
-        assert!(js.contains("document.querySelector(\"#submit-btn\")"));
+        // CSS is now tried via a `sel` variable, with a visible-text fallback
+        // appended (issue #24-B). It must still use querySelector (not xpath).
+        assert!(js.contains("const sel = \"#submit-btn\""));
+        assert!(js.contains("document.querySelector(sel)"));
         assert!(!js.contains("document.evaluate"));
     }
 
+    #[test]
+    fn test_build_find_element_js_text_fallback() {
+        // A bare label gets a text-matching fallback so `click "購入手続きへ"`
+        // resolves by visible text, not just CSS (issue #24-B).
+        let js = build_find_element_js("購入手続きへ");
+        assert!(js.contains("購入手続きへ"));
+        assert!(js.contains("interactive")); // the text-match branch
+        assert!(js.contains("textOf"));
+        // `text=` forces the text path (skips CSS).
+        let forced = build_find_element_js("text=Buy now");
+        assert!(forced.contains("true ? null")); // force_text => css skipped
+                                                 // xpath is unchanged.
+        let xp = build_find_element_js("xpath=//button");
+        assert!(xp.contains("document.evaluate"));
+        assert!(!xp.contains("interactive"));
+    }
+
     #[test]
     fn test_build_selector_js_xpath() {
         let js = build_selector_js("xpath=//button[@id='ok']");

cli/src/native/interaction.rs

@@ -306,32 +306,50 @@ pub async fn fill(
     )
     .await?;
 
-    // Focus the element
-    client
-        .send_command_typed::<_, Value>(
-            "Runtime.callFunctionOn",
-            &CallFunctionOnParams {
-                function_declaration: "function() { this.focus(); }".to_string(),
-                object_id: Some(object_id.clone()),
-                arguments: None,
-                return_by_value: Some(true),
-                await_promise: Some(false),
-            },
-            Some(&effective_session_id),
-        )
-        .await?;
+    // Emulate a real edit so framework-controlled inputs (React/Vue) and
+    // site-side listeners actually see the change (issue #25): the old path set
+    // `this.value` directly and used Input.insertText, which left React's
+    // internal value-tracker out of sync and never fired change/blur — so
+    // dependent logic (e.g. Mercari's postal-code → 都道府県 autocomplete) never
+    // ran even though the value was visible. Set the value through the element's
+    // PROTOTYPE setter (which React's _valueTracker hooks), then dispatch
+    // input → change → blur/focusout. `type <sel> <text>` remains for sites that
+    // need per-keystroke events.
+    let fill_js = format!(
+        r#"function() {{
+            const el = this;
+            const v = {val};
+            try {{ el.focus(); }} catch (e) {{}}
+            const tag = el.tagName;
+            const fire = (type, ctor) => el.dispatchEvent(new (ctor || Event)(type, {{ bubbles: true }}));
+            if (tag === 'SELECT') {{
+                el.value = v; fire('input'); fire('change'); return true;
+            }}
+            if (el.isContentEditable) {{
+                el.textContent = v; fire('input', window.InputEvent || Event); fire('change');
+                try {{ el.blur(); }} catch (e) {{}} fire('focusout'); return true;
+            }}
+            const proto = tag === 'TEXTAREA' ? window.HTMLTextAreaElement.prototype
+                                             : window.HTMLInputElement.prototype;
+            const desc = Object.getOwnPropertyDescriptor(proto, 'value');
+            const set = desc && desc.set ? (x) => desc.set.call(el, x) : (x) => {{ el.value = x; }};
+            set('');                                   // reset the framework tracker
+            fire('input', window.InputEvent || Event);
+            set(v);                                    // native setter → React/Vue registers
+            fire('input', window.InputEvent || Event);
+            fire('change');
+            try {{ el.blur(); }} catch (e) {{}}
+            fire('focusout');                          // blur-triggered lookups/validation
+            return true;
+        }}"#,
+        val = serde_json::to_string(value).unwrap_or_default()
+    );
 
-    // Select all + delete to clear
     client
         .send_command_typed::<_, Value>(
             "Runtime.callFunctionOn",
             &CallFunctionOnParams {
-                function_declaration: r#"function() {
-                    this.select && this.select();
-                    this.value = '';
-                    this.dispatchEvent(new Event('input', { bubbles: true }));
-                }"#
-                .to_string(),
+                function_declaration: fill_js,
                 object_id: Some(object_id),
                 arguments: None,
                 return_by_value: Some(true),
@@ -341,17 +359,6 @@ pub async fn fill(
         )
         .await?;
 
-    // Insert text (keyboard input dispatched at page level, use parent session_id)
-    client
-        .send_command_typed::<_, Value>(
-            "Input.insertText",
-            &InsertTextParams {
-                text: value.to_string(),
-            },
-            Some(session_id),
-        )
-        .await?;
-
     Ok(())
 }
 
@@ -557,6 +564,48 @@ pub async fn press_key_with_modifiers(
     Ok(())
 }
 
+/// Dispatch a SINGLE key event (`keyDown` or `keyUp`) carrying the full key
+/// descriptor — `key`, `code`, `windowsVirtualKeyCode`/`nativeVirtualKeyCode`,
+/// and (on key-down) printable `text`. Powers the `keydown`/`keyup` commands.
+///
+/// The previous implementation sent only `{key}`, so games and shortcut handlers
+/// that read `event.code` (e.g. `"KeyD"`, `"ArrowRight"`) or `event.keyCode` saw
+/// nothing — a held key set no movement flag and did nothing (dogfood: holding a
+/// direction in a canvas platformer barely nudged the player). Sending the same
+/// descriptor `press` uses makes hold-to-move work regardless of which field the
+/// page keys off.
+pub async fn dispatch_single_key(
+    client: &CdpClient,
+    session_id: &str,
+    key: &str,
+    event_type: &str,
+) -> Result<(), String> {
+    let (key_name, code, key_code) = named_key_info(key);
+    // Printable text is only meaningful on key-down; key-up never inserts.
+    let text = if event_type == "keyDown" {
+        key_text(&key_name)
+    } else {
+        None
+    };
+    client
+        .send_command_typed::<_, Value>(
+            "Input.dispatchKeyEvent",
+            &DispatchKeyEventParams {
+                event_type: event_type.to_string(),
+                key: Some(key_name),
+                code: Some(code),
+                text: text.clone(),
+                unmodified_text: text,
+                windows_virtual_key_code: Some(key_code),
+                native_virtual_key_code: Some(key_code),
+                modifiers: None,
+            },
+            Some(session_id),
+        )
+        .await?;
+    Ok(())
+}
+
 pub async fn scroll(
     client: &CdpClient,
     session_id: &str,

cli/src/native/screenshot.rs

@@ -60,6 +60,9 @@ pub struct ScreenshotOptions {
     pub quality: Option<i32>,
     pub annotate: bool,
     pub output_dir: Option<String>,
+    /// Explicit pixel region (x, y, width, height) — `--clip` (issue #34). Takes
+    /// precedence over selector/full_page.
+    pub clip: Option<(f64, f64, f64, f64)>,
 }
 
 impl Default for ScreenshotOptions {
@@ -72,6 +75,7 @@ impl Default for ScreenshotOptions {
             quality: None,
             annotate: false,
             output_dir: None,
+            clip: None,
         }
     }
 }
@@ -187,7 +191,16 @@ async fn capture_screenshot_base64(
         capture_beyond_viewport: if options.full_page { Some(true) } else { None },
     };
 
-    if options.full_page {
+    if let Some((x, y, width, height)) = options.clip {
+        // Explicit pixel region wins over selector/full_page (issue #34).
+        params.clip = Some(Viewport {
+            x,
+            y,
+            width,
+            height,
+            scale: 1.0,
+        });
+    } else if options.full_page {
         let metrics: Value = client
             .send_command_no_params("Page.getLayoutMetrics", Some(session_id))
             .await?;

cli/src/output.rs

@@ -186,6 +186,78 @@ pub fn print_response_with_opts(resp: &Response, action: Option<&str>, opts: &Ou
     }
 
     if let Some(data) = &resp.data {
+        // A click that opened a new tab: surface it so the agent doesn't read the
+        // unchanged old page as a failed click (issue #24-A).
+        if let Some(opened) = data.get("openedTab") {
+            let tid = opened.get("tabId").and_then(|v| v.as_str()).unwrap_or("?");
+            let url = opened.get("url").and_then(|v| v.as_str()).unwrap_or("");
+            let followed = data
+                .get("followed")
+                .and_then(|v| v.as_bool())
+                .unwrap_or(false);
+            let verb = if followed {
+                "switched to new tab"
+            } else {
+                "opened new tab"
+            };
+            eprintln!(
+                "{} {} [{}] {}",
+                color::cyan("→"),
+                verb,
+                tid,
+                color::dim(url)
+            );
+        }
+
+        // `current`: the active tab's stable handle (#26).
+        if data
+            .get("current")
+            .and_then(|v| v.as_bool())
+            .unwrap_or(false)
+        {
+            let tid = data.get("tabId").and_then(|v| v.as_str()).unwrap_or("?");
+            let title = data.get("title").and_then(|v| v.as_str()).unwrap_or("");
+            let url = data.get("url").and_then(|v| v.as_str()).unwrap_or("");
+            let target = data.get("targetId").and_then(|v| v.as_str()).unwrap_or("");
+            println!("{} [{}] {} - {}", color::cyan("→"), tid, title, url);
+            println!("      {}", color::dim(&format!("target: {}", target)));
+            return;
+        }
+
+        // Cloudflare challenge/clearance preflight (`cf-status`). Checked early
+        // because its response carries `url`/`title`, which later generic
+        // renderers would otherwise swallow.
+        if action == Some("cf_status") {
+            let challenged = data.get("challenged").and_then(|v| v.as_bool()).unwrap_or(false);
+            let rec = data.get("recommendation").and_then(|v| v.as_str()).unwrap_or("?");
+            let cl = data.get("clearance");
+            let present = cl.and_then(|c| c.get("present")).and_then(|v| v.as_bool()).unwrap_or(false);
+            let expired = cl.and_then(|c| c.get("expired")).and_then(|v| v.as_bool()).unwrap_or(false);
+            let expires_in = cl.and_then(|c| c.get("expiresIn")).and_then(|v| v.as_i64());
+            let device = data.get("deviceVerified").and_then(|v| v.as_bool()).unwrap_or(false);
+
+            let (icon, headline) = match rec {
+                "proceed" => (color::success_indicator().to_string(), "cleared — no challenge, proceed"),
+                "solve" => (color::warning_indicator().to_string(), "Cloudflare challenge active, no valid clearance — solve it"),
+                "reissue" => (color::warning_indicator().to_string(), "challenge active but a clearance cookie exists — stale (IP/UA changed?), re-solve"),
+                _ => (color::cyan("•").to_string(), "unknown"),
+            };
+            println!("{} {}", icon, headline);
+            println!("  challenged:     {}", if challenged { "yes" } else { "no" });
+            let cl_desc = if !present {
+                "absent".to_string()
+            } else if expired {
+                "present but EXPIRED".to_string()
+            } else if let Some(s) = expires_in {
+                format!("valid, expires in {}m {}s", s / 60, s % 60)
+            } else {
+                "present (session)".to_string()
+            };
+            println!("  cf_clearance:   {}", cl_desc);
+            println!("  device trusted: {}", if device { "yes (CF_VERIFIED_DEVICE)" } else { "no" });
+            return;
+        }
+
         // Dialog status response
         if action == Some("dialog") {
             if let Some(has_dialog) = data.get("hasDialog").and_then(|v| v.as_bool()) {
@@ -297,6 +369,39 @@ pub fn print_response_with_opts(resp: &Response, action: Option<&str>, opts: &Ou
         // Snapshot
         if let Some(snapshot) = data.get("snapshot").and_then(|v| v.as_str()) {
             print_with_boundaries(snapshot, origin, opts);
+            // Canvas-app hint: the tree was near-empty but the page paints to a
+            // <canvas>, so refs are a dead end — point at the screenshot path.
+            if let Some(note) = data.get("note").and_then(|v| v.as_str()) {
+                eprintln!("{}", color::dim(note));
+            }
+            return;
+        }
+        // Frame list (`chrome-use frames`)
+        if action == Some("frames") {
+            if let Some(list) = data.get("frames").and_then(|v| v.as_array()) {
+                let count = list.len();
+                println!(
+                    "{}",
+                    color::bold(&format!("{} frame{}", count, if count == 1 { "" } else { "s" }))
+                );
+                for f in list {
+                    let idx = f.get("index").and_then(|v| v.as_i64()).unwrap_or(0);
+                    let kind = f.get("kind").and_then(|v| v.as_str()).unwrap_or("?");
+                    let url = f.get("url").and_then(|v| v.as_str()).unwrap_or("");
+                    let len = f.get("textLen").and_then(|v| v.as_i64()).unwrap_or(0);
+                    println!(
+                        "  [{}] {:<6} {} chars  {}",
+                        idx,
+                        kind,
+                        len,
+                        color::dim(if url.is_empty() { "(about:blank)" } else { url })
+                    );
+                }
+                eprintln!(
+                    "{}",
+                    color::dim("read everything with: chrome-use get text --all-frames")
+                );
+            }
             return;
         }
         // Title
@@ -526,19 +631,21 @@ pub fn print_response_with_opts(resp: &Response, action: Option<&str>, opts: &Ou
         // Tab switch
         if action == Some("tab_switch") {
             if let Some(tab_id) = data.get("tabId").and_then(|v| v.as_str()) {
-                if let Some(url) = data.get("url").and_then(|v| v.as_str()) {
-                    println!(
-                        "{} Switched to tab [{}] ({})",
-                        color::success_indicator(),
-                        tab_id,
-                        url
-                    );
+                let warning = data.get("warning").and_then(|v| v.as_str());
+                // A non-responding session isn't a real success — show a warning
+                // indicator instead of the green ✓ (issue #29.3).
+                let indicator = if warning.is_some() {
+                    color::warning_indicator()
                 } else {
-                    println!(
-                        "{} Switched to tab [{}]",
-                        color::success_indicator(),
-                        tab_id
-                    );
+                    color::success_indicator()
+                };
+                if let Some(url) = data.get("url").and_then(|v| v.as_str()) {
+                    println!("{} Switched to tab [{}] ({})", indicator, tab_id, url);
+                } else {
+                    println!("{} Switched to tab [{}]", indicator, tab_id);
+                }
+                if let Some(w) = warning {
+                    eprintln!("{}", color::dim(w));
                 }
                 return;
             }
@@ -1725,6 +1832,8 @@ Pass --hide-scrollbars false when launching to keep native scrollbars visible.
 
 Options:
   --full, -f           Capture full page (not just viewport)
+  [selector]           Capture just an element (CSS or @ref), e.g. `screenshot ".header" h.png`
+  --clip <x,y,w,h>     Capture a pixel region, e.g. `screenshot --clip 0,0,200,40 corner.png`
   --annotate           Overlay numbered labels on interactive elements.
                        Each label [N] corresponds to ref @eN from snapshot.
                        Prints a legend mapping labels to element roles/names.
@@ -1745,6 +1854,8 @@ Examples:
   chrome-use screenshot
   chrome-use screenshot ./screenshot.png
   chrome-use screenshot --full ./full-page.png
+  chrome-use screenshot ".header .indicator" corner.png  # just one element
+  chrome-use screenshot --clip 1600,0,200,40 corner.png  # a pixel region
   chrome-use screenshot --annotate              # Labeled screenshot + legend
   chrome-use screenshot --annotate ./page.png   # Save annotated screenshot
   chrome-use screenshot --annotate --json       # JSON output with annotations
@@ -1886,7 +1997,9 @@ Usage: chrome-use get <subcommand> [args]
 Retrieves various types of information from elements or the page.
 
 Subcommands:
-  text <selector>            Get text content of element
+  text [selector]            Element text; no selector = WHOLE PAGE, all frames
+  text --main                Main-content text only (skip nav/header/sidebar)
+  text --pierce              Read through CLOSED shadow DOM (injected panels)
   html <selector>            Get inner HTML of element
   value <selector>           Get value of input element
   attr <selector> <name>     Get attribute value
@@ -1902,7 +2015,10 @@ Global Options:
   --session <name>     Use specific session
 
 Examples:
-  chrome-use get text @e1
+  chrome-use get text                  # whole page across ALL frames (default)
+  chrome-use get text @e1              # one element
+  chrome-use get text --main           # main content, no nav/sidebar boilerplate
+  chrome-use frames                    # list frames + where the text lives
   chrome-use get html "#content"
   chrome-use get value "#email-input"
   chrome-use get attr "#link" href
@@ -2773,6 +2889,20 @@ Notes:
   - Streaming is always enabled. Set AGENT_BROWSER_STREAM_PORT to bind to a
     specific port instead of the default OS-assigned port.
 
+The WS is BIDIRECTIONAL — the high-throughput way to drive a live/real-time page
+(games, canvas apps) instead of one screenshot + one CLI call per action:
+  - Server -> client (JSON text frames):
+      {"type":"frame","data":"<base64 jpeg>"}   live screencast (~60fps)
+      plus status / tabs messages.
+  - Client -> server (send JSON text):
+      {"type":"input_keyboard","eventType":"keyDown|keyUp","key":" ","code":"Space",
+       "windowsVirtualKeyCode":32}
+      {"type":"input_mouse","eventType":"mousePressed|mouseReleased|mouseMoved",
+       "x":640,"y":360,"button":"left","clickCount":1}
+      {"type":"input_touch","eventType":"touchStart|touchEnd","touchPoints":[...]}
+  Connect once and run a tight local loop: read frames, send timed input — no
+  per-action process spawn, no round-trip. Works over the extension relay too.
+
 Global Options:
   --json               Output as JSON
   --session <name>     Use specific session
@@ -3065,7 +3195,12 @@ Core Commands:
   dblclick <sel>             Double-click element
   type <sel> <text>          Type into element
   fill <sel> <text>          Clear and fill
-  press <key>                Press key (Enter, Tab, Control+a)
+  press <key> [--hold <ms>]  Press key (Enter, Tab, Control+a). --hold keeps it
+                             down <ms> then releases — precise (in-daemon), for
+                             games/charge: `press d --hold 800`
+  keydown <key>              Hold a key down (no auto-release) — for games/shortcuts
+  keyup <key>                Release a held key. Pair with keydown to hold-to-move:
+                             `keydown d` … `keyup d`
   keyboard type <text>       Type text with real keystrokes (no selector)
   keyboard inserttext <text> Insert text without key events
   hover <sel>                Hover element
@@ -3093,10 +3228,15 @@ Navigation:
 
 Get Info:  chrome-use get <what> [selector]
   text, html, value, attr <name>, title, url, count, box, styles, cdp-url
+  text (no selector = whole page, all frames), text --main, frames (list)
 
 Check State:  chrome-use is <what> <selector>
   visible, enabled, checked
 
+Anti-bot:  chrome-use stealth | cf-status
+  stealth         stealth self-check (webdriver/UA/plugins + overrides)
+  cf-status       Cloudflare challenge + cf_clearance preflight (skip re-solving)
+
 Find Elements:  chrome-use find <locator> <value> <action> [text]
   role, text, label, placeholder, alt, title, testid, first, last, nth
 
@@ -3191,6 +3331,7 @@ Confirmation:
 Sessions:
   session                    Show current session name
   session list               List active sessions
+  sessions                   List running session daemons (alias of daemon status)
   daemon status              List running session daemons (+ relay state)
   daemon restart             Kill all session daemons; keeps the extension relay
                              up. Clears stale/cross-leaked state after an upgrade.

cli/src/upgrade.rs

@@ -52,6 +52,27 @@ fn is_newer(latest: &str, current: &str) -> bool {
     matches!((parse_version(latest), parse_version(current)), (Some(l), Some(c)) if l > c)
 }
 
+/// Public semver-ish comparison (`latest` strictly newer than `current`), so
+/// `doctor` can flag a stale extension/CLI without re-implementing parsing.
+pub fn version_is_newer(latest: &str, current: &str) -> bool {
+    is_newer(latest, current)
+}
+
+/// The latest CLI version recorded by the background update check, if any.
+/// `doctor` uses it to show "a newer chrome-use is available" without a network
+/// call (the `__update-check` worker refreshes the cache out of band).
+pub fn cached_latest_version() -> Option<String> {
+    std::fs::read_to_string(update_cache_path())
+        .ok()
+        .and_then(|s| serde_json::from_str::<serde_json::Value>(&s).ok())
+        .and_then(|j| {
+            j.get("latest")
+                .and_then(|v| v.as_str())
+                .map(|s| s.to_string())
+        })
+        .filter(|s| !s.is_empty())
+}
+
 /// Hidden `__update-check` subcommand: fetch the latest release tag and cache it.
 /// Spawned detached by [`maybe_notify_update`] so the network call never blocks a
 /// real command. Uses `curl` (no extra deps, matches `upgrade`).

extensions/ab-connect/background.js

@@ -30,6 +30,16 @@ const tabs = new Map()
 const sessionToTab = new Map()
 /** child (OOPIF/worker) sessionId -> tabId */
 const childSessionToTab = new Map()
+/** sessionId -> CDP targetId, kept ACROSS detach so a dead `cb-tab-<oldTabId>`
+ * session can be recovered by its stable targetId when the cross-process nav
+ * gave the tab a new Chrome tabId (issue #24). Capped to bound memory. */
+const sessionTargets = new Map()
+function rememberSessionTarget(sessionId, targetId) {
+  if (!sessionId || !targetId) return
+  sessionTargets.delete(sessionId)
+  sessionTargets.set(sessionId, targetId)
+  if (sessionTargets.size > 256) sessionTargets.delete(sessionTargets.keys().next().value)
+}
 /** tab-group name -> chrome tabGroups id (best-effort cache) */
 const groupIdByName = new Map()
 
@@ -108,6 +118,12 @@ function connectHost() {
     // reconnect. Keep chrome.debugger attached so reconnect is cheap.
     for (const tabId of tabs.keys()) setBadge(tabId, 'connecting')
   })
+  // Report our version so the host can tell the CLI/`doctor` which extension
+  // build is live (otherwise the extension version is a black box — the user
+  // can't tell they're on an old one). Best-effort; ignored by older hosts.
+  try {
+    postToHost({ method: 'hello', version: chrome.runtime.getManifest().version })
+  } catch {}
   // Tell the daemon about everything we already have attached, then attach
   // anything new.
   reannounceAttachedTabs()
@@ -149,23 +165,88 @@ function tabForTarget(targetId) {
   return null
 }
 
-// Best-effort recovery for a stale `cb-tab-<tabId>` session: the handle is gone
-// from our maps, but if the underlying Chrome tab still exists and is eligible,
-// re-attach to it and return its id so the in-flight command can be retried.
-// Returns null when the tab is genuinely gone (closed / restricted), in which
-// case the caller surfaces the stale-session error. (issue #20.1)
+// The STABLE Chrome tabId encoded in a `cb-tab-<tabId>` session id (#17), or
+// null for any other session shape (child/iframe sessions). The tabId is the
+// real source of truth: it survives the renderer-process swaps (cross-origin
+// OAuth/SSO navs) that tear down the page's CDP target — which is why binding to
+// it (like claude-in-chrome) rides through the hop that killed the old
+// target/sessionId binding (issue #23).
+function tabIdFromSession(sessionId) {
+  const m = /^cb-tab-(\d+)$/.exec(sessionId || '')
+  return m ? Number(m[1]) : null
+}
+
+// Ensure the debugger is attached to a `cb-tab-<tabId>` session's tab, re-attaching
+// across the transient window of a process swap (with a couple of short retries).
+// Returns the tabId on success, or null when the tab is genuinely gone
+// (closed / restricted). (issues #20.1, #23)
 async function recoverSessionTab(sessionId) {
-  const m = /^cb-tab-(\d+)$/.exec(sessionId)
-  if (!m) return null
-  const tabId = Number(m[1])
-  const tab = await chrome.tabs.get(tabId).catch(() => null)
-  if (!eligible(tab)) return null
-  try {
-    await attachTab(tabId)
-  } catch {
-    return null
+  const tabId = tabIdFromSession(sessionId)
+  // 1) Fast path: the encoded Chrome tabId still exists — re-attach it (covers
+  //    the common renderer-process swap where the tabId is preserved, #23).
+  if (tabId != null) {
+    for (let i = 0; i < 3; i++) {
+      const tab = await chrome.tabs.get(tabId).catch(() => null)
+      if (!eligible(tab)) break // tabId is gone — fall through to targetId recovery
+      try {
+        await attachTab(tabId)
+        if (tabs.has(tabId)) return tabId
+      } catch {
+        // mid-swap: tab exists but isn't attachable yet — back off and retry.
+      }
+      await new Promise((r) => setTimeout(r, 120 + i * 150))
+    }
+  }
+  // 2) The Chrome tabId is gone, but the CDP targetId is STABLE across the nav.
+  //    Some cross-process hops (Mercari's signin token exchange) give the tab a
+  //    NEW tabId while keeping the same target, so `cb-tab-<oldTabId>` can't be
+  //    recovered by tabId. Find the tab now hosting our remembered targetId via
+  //    chrome.debugger.getTargets(), attach it, and ALIAS the dead session to it
+  //    so the daemon's session id keeps resolving. Longer window: this hop can
+  //    take several seconds to settle (issue #24).
+  const targetId = sessionTargets.get(sessionId)
+  if (targetId) {
+    for (let i = 0; i < 6; i++) {
+      const targets = await chrome.debugger.getTargets().catch(() => null)
+      const t = targets && targets.find((x) => x.id === targetId && x.tabId != null)
+      if (t && t.tabId != null) {
+        const tab = await chrome.tabs.get(t.tabId).catch(() => null)
+        if (eligible(tab)) {
+          try {
+            await attachTab(t.tabId)
+            if (tabs.has(t.tabId)) {
+              sessionToTab.set(sessionId, t.tabId) // alias dead session -> live tab
+              return t.tabId
+            }
+          } catch {
+            // not attachable yet — keep waiting for the swap to settle.
+          }
+        }
+      }
+      await new Promise((r) => setTimeout(r, 300 + i * 300))
+    }
+  }
+  return null
+}
+
+// Send a CDP command to a tab, riding a debugger detach that can happen between
+// our attach check and the command itself (a renderer-process swap mid-flight).
+// On a detached-style failure, drop the stale handle, re-attach the stable tab,
+// and retry once — so a cross-process nav never surfaces as a hard error (#23).
+async function sendCdpToTab(tabId, method, params) {
+  const dbg = { tabId }
+  try {
+    return await chrome.debugger.sendCommand(dbg, method, params)
+  } catch (e) {
+    const msg = String((e && e.message) || e)
+    if (!/detached|not attached|target.*(closed|gone)|no target|cannot access|frame.*detached/i.test(msg)) {
+      throw e
+    }
+    detachTab(tabId, false)
+    const ok = await recoverSessionTab(`cb-tab-${tabId}`)
+    if (!ok) throw e
+    return await chrome.debugger.sendCommand(dbg, method, params)
   }
-  return tabs.has(tabId) ? tabId : null
 }
 
 function anyConnectedTab() {
@@ -226,24 +307,26 @@ async function handleForwardCdpCommand(msg) {
   // Fail loudly instead so the agent sees an actionable error, not bad data.
   let tabId
   if (sessionId) {
-    tabId = tabForSession(sessionId)
-    if (!tabId) {
-      // The session's debugger handle is gone, but `cb-tab-<tabId>` encodes the
-      // STABLE Chrome tabId (#17). A cross-process navigation (e.g. an SSO
-      // redirect to another origin), a service-worker restart, or DevTools
-      // briefly stealing the debugger all tear the handle down while the tab
-      // itself lives on. Before failing, try to transparently re-attach to that
-      // same tab and retry — so `open`/`navigate`/`eval` self-heal instead of
-      // dead-ending the agent (issue #20.1). attachTab re-mints the identical
-      // `cb-tab-<tabId>` session, so the daemon's binding stays valid.
-      tabId = await recoverSessionTab(sessionId)
-      if (!tabId) {
+    // The stable Chrome tabId encoded in `cb-tab-<tabId>` is the source of truth
+    // (it survives renderer-process swaps; the CDP target/sessionId does not).
+    // Resolve via it primarily — don't depend on a session→tab map entry that the
+    // detach handler may have cleared — and ensure the debugger is attached,
+    // re-attaching across a cross-process nav before failing (issues #20.1, #23).
+    // `tabForSession` still covers child/iframe sessions that aren't `cb-tab-*`.
+    tabId = tabIdFromSession(sessionId) ?? tabForSession(sessionId)
+    if (tabId == null) {
+      throw new Error(`unknown sessionId ${sessionId} for ${method}`)
+    }
+    if (!tabs.has(tabId)) {
+      const recovered = await recoverSessionTab(sessionId)
+      if (!recovered) {
         throw new Error(
           `stale sessionId ${sessionId} for ${method}: its tab is gone (closed, ` +
             `navigated across processes, or lost after an extension restart). ` +
             `Re-attach by re-opening your target URL before retrying.`,
         )
       }
+      tabId = recovered
     }
   } else if (typeof params?.targetId === 'string') {
     tabId = tabForTarget(params.targetId)
@@ -253,18 +336,17 @@ async function handleForwardCdpCommand(msg) {
     // applies to any attached tab.
     tabId = anyConnectedTab()
   }
-  if (!tabId) throw new Error(`no attached tab for ${method}`)
-  const dbg = { tabId }
+  if (tabId == null) throw new Error(`no attached tab for ${method}`)
 
   // Re-enabling Runtime can leave a stale state; bounce it (matches upstream).
   if (method === 'Runtime.enable') {
     try {
-      await chrome.debugger.sendCommand(dbg, 'Runtime.disable')
+      await sendCdpToTab(tabId, 'Runtime.disable', undefined)
       await new Promise((r) => setTimeout(r, 30))
     } catch {}
-    return await chrome.debugger.sendCommand(dbg, 'Runtime.enable', params)
+    return await sendCdpToTab(tabId, 'Runtime.enable', params)
   }
-  return await chrome.debugger.sendCommand(dbg, method, params)
+  return await sendCdpToTab(tabId, method, params)
 }
 
 // ---- attach / detach ------------------------------------------------------
@@ -303,6 +385,7 @@ async function attachTab(tabId) {
   const entry = { sessionId, targetId }
   tabs.set(tabId, entry)
   sessionToTab.set(sessionId, tabId)
+  rememberSessionTarget(sessionId, targetId)
   setBadge(tabId, port ? 'on' : 'connecting')
   postToHost({
     method: 'forwardCDPEvent',

extensions/ab-connect/manifest.json

@@ -1,7 +1,7 @@
 {
   "manifest_version": 3,
   "name": "chrome-use",
-  "version": "0.4.6",
+  "version": "0.4.9",
   "description": "Let chrome-use drive your logged-in Chrome \u2014 install once, no token, no per-use confirmation.",
   "key": "MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEA6vQIyscGIPYPZdSpPwPL0+0gxUROyRgCpmvCSDoc8XUm4qm97VbKnD9Ijc1lV22lNWZtE78gaRjt6BeSfuMgnBymnhLKjN1gU6AI5QUU0mrJyeHdWKvrKQR5FmsM2A7Xr1ykE2SiiS8zNUS3Y/6O5l+Nva7wrVy6E4a2dkBVQkOsu+DV+nEZvhIyuDY5D5SPXqNwUTWTaglwj5mjvHz36xSwCWlPmrtJ+ED0AUyrb2z4GIOmvk4kqtBVrh/UD058klLo4CkYOnIybB5aV6WYuwarfPY4bF/dLggPem+ewLNTUNBuwrxj/A4nUv0LJTuRO8rR7f8WR9qnRCY0Ic5saQIDAQAB",
   "icons": {

package.json

@@ -1,6 +1,6 @@
 {
   "name": "chrome-use",
-  "version": "1.3.0",
+  "version": "1.5.9",
   "description": "chrome-use — drive your real, logged-in Chrome from any AI agent, stealth by default",
   "type": "module",
   "packageManager": "pnpm@11.1.3",

skill-data/core/SKILL.md

@@ -127,6 +127,19 @@ cadence, and scroll/drag ease. Default `off`; a per-navigation detector
 auto-escalates pages guarded by Akamai/PerimeterX/DataDome to `human`. Leave it
 on auto; force `human` only when you already know the target scores behaviour.
 
+**Cloudflare clearance — solve once, reuse.** Passing a Cloudflare challenge
+mints a `cf_clearance` cookie (HttpOnly — invisible to `eval`/`document.cookie`;
+read it via `chrome-use cookies`). It's bound to your **IP + User-Agent**: reuse
+the same exit IP and UA and you skip the challenge until it expires. Driving the
+user's real Chrome (relay) persists it natively; for isolated sessions,
+`--session-name <name>` save/restores it. Before spending effort solving, run
+`chrome-use cf-status` (aliases `cf`, `clearance`): it reports whether the page
+is *currently* a Cloudflare challenge and whether a still-valid `cf_clearance`
+exists, with a recommendation — `proceed` (already cleared, don't re-solve),
+`solve` (challenge up, no clearance), or `reissue` (clearance present but page
+still blocks → IP/UA drifted, re-solve). Use it as a preflight to avoid
+re-solving what you already cleared.
+
 ## Two ways to drive a page — and when to drop to `eval`
 
 You have a **real Chrome with the user's DOM**. Two layers, mix them freely:
@@ -207,7 +220,11 @@ assigned fresh on every snapshot.
 For unstructured reading (no refs needed):
 
 ```bash
-chrome-use get text @e1                # visible text of an element
+chrome-use get text                    # WHOLE PAGE — all frames by default (see below)
+chrome-use get text @e1                # visible text of one element (or a CSS selector)
+chrome-use get text --main             # main content only — skip nav/header/sidebar
+chrome-use get text --pierce           # read through CLOSED shadow DOM (injected panels)
+chrome-use frames                      # list every frame + where the text lives
 chrome-use get html @e1                # innerHTML
 chrome-use get attr @e1 href           # any attribute
 chrome-use get value @e1               # input value
@@ -216,6 +233,27 @@ chrome-use get url                     # current URL
 chrome-use get count ".item"           # count matching elements
 ```
 
+**Whole-page text is cross-frame by default.** `chrome-use get text` with no
+selector aggregates visible text across **every** frame — top document plus
+same-process child frames plus cross-origin iframes — so you never silently miss
+content that lives in an iframe (Yahoo Auctions / Rakuten / Mercari shop
+descriptions, embedded checkout/spec frames). Each child frame is delimited with
+a `----- frame [kind] url -----` marker. You do **not** need to remember a flag —
+the default already reads all frames. (`--all-frames` is still accepted as an
+explicit alias.)
+
+So: when text looks missing or wrong, you don't have to guess — just
+`chrome-use get text` reads everything. To **see** the structure (which frame
+holds what), run `chrome-use frames`. To **cut boilerplate** (global nav/header/
+footer, "related items" sidebars), use `chrome-use get text --main`. If content
+is lazy-loaded, `scroll` it into view first, then read.
+
+**Closed shadow DOM.** Some injected UI (browser-extension debug panels, web
+components) renders into a *closed* shadow root that `eval`/`innerText` cannot
+read. `chrome-use get text --pierce` reads through closed shadow roots and child
+documents via the CDP DOM tree — use it when content is clearly on screen (you
+see it in a screenshot) but `get text`/`eval` come back empty.
+
 ## Interacting
 
 ```bash
@@ -226,8 +264,11 @@ chrome-use hover @e1                   # hover
 chrome-use focus @e1                   # focus (useful before keyboard input)
 chrome-use fill @e2 "hello"            # clear then type
 chrome-use type @e2 " world"           # type without clearing
-chrome-use press Enter                 # press a key at current focus
+chrome-use press Enter                 # press a key at current focus (down+up)
 chrome-use press Control+a             # key combination
+chrome-use keydown d                   # HOLD a key down (no auto-release)
+chrome-use keyup d                     # release it — pair them to hold-to-move
+                                          # in a game: `keydown d; sleep; keyup d`
 chrome-use check @e3                   # check checkbox
 chrome-use uncheck @e3                 # uncheck
 chrome-use select @e4 "option-value"   # native <select> only
@@ -240,11 +281,11 @@ chrome-use pick @e4 --option "Europe"  # ANY combobox (react-select / ARIA /
                                           # (no silent no-op). Use this for custom
                                           # dropdowns where `select` returns ✓ but
                                           # changes nothing.
-chrome-use upload @e5 file1.pdf        # upload file(s) — NOTE: needs a --launch/direct-CDP
-                                          # session. Over the extension relay it CANNOT work
-                                          # (Chrome's chrome.debugger forbids it); chrome-use
-                                          # errors with a hint. Carry your login into a launched
-                                          # session via `cookies export` | `cookies set --curl`.
+chrome-use upload @e5 file1.pdf        # upload file(s) — works over the extension relay too:
+                                          # chrome.debugger forbids setFileInputFiles, so the
+                                          # file's bytes are streamed into the page and rebuilt as
+                                          # a File there (chunked under native-messaging's 1 MiB cap).
+                                          # Works on file <input>s and drop/paste composers (e.g. X).
 chrome-use scroll down 500             # scroll page (up/down/left/right)
 chrome-use scrollintoview @e1          # scroll element into view
 chrome-use drag @e1 @e2                # drag and drop
@@ -296,6 +337,54 @@ chrome-use click --coords 449,320   # same, explicit flag
 
 A bare-number argument is always a coordinate, never a selector.
 
+### Canvas / WebGL apps (games, map & 3D viewers, drawing tools)
+
+These paint everything to a `<canvas>` and expose **almost no accessibility
+tree**, so `snapshot` comes back near-empty and refs are a dead end. `snapshot`
+detects this and prints a one-line hint. Drive them the screenshot way:
+
+```bash
+chrome-use screenshot /tmp/s.png       # SEE the state (your only read path —
+                                          # eval/get text return nothing useful)
+chrome-use click 640 360               # interact by viewport coordinate
+chrome-use press d --hold 800          # hold-to-move, precise (timed in-daemon —
+                                          # NOT keydown+shell-sleep+keyup, which
+                                          # adds ~250ms jitter per round-trip)
+chrome-use press Space                 # discrete actions (jump/attack/confirm)
+```
+
+**Don't drive frame-by-frame with one CLI call per action** — that's the slowest,
+lowest-fidelity way (each call is a process spawn + round-trip). Script a *timed
+sequence in a single round-trip* with `batch` (it sends each step to the running
+daemon; `press --hold` and `wait` block in-daemon, so timing is precise):
+
+```bash
+chrome-use batch "press d --hold 900" "press j" "press j" "wait 200" "press d --hold 500"
+```
+
+Also try reading real state instead of pixels: `eval` runs in the page's main
+world, so for a framework/engine game you can often reach its globals (e.g. a
+Phaser/PIXI/Three instance, a store, `window.__GAME__`) and read positions/score
+directly — far better than guessing from a screenshot.
+
+**For genuinely real-time driving, drop the CLI entirely and use the WebSocket.**
+`chrome-use stream enable` opens a bidirectional WS (`stream status` prints the
+`ws://127.0.0.1:<port>`). Connect once and you get a live ~60fps screencast AND
+can send input on the same socket — no per-action process spawn, no round-trip,
+works over the extension relay:
+
+```js
+// node (global WebSocket): live frames + locally-timed input
+const ws = new WebSocket("ws://127.0.0.1:PORT")
+ws.onmessage = e => { const m = JSON.parse(e.data); if (m.type==="frame") {/* base64 jpeg */} }
+const k = (eventType,key,code,vk) => ws.send(JSON.stringify({type:"input_keyboard",eventType,key,code,windowsVirtualKeyCode:vk}))
+k("keyDown"," ","Space",32); setTimeout(()=>k("keyUp"," ","Space",32), 80)   // a jump
+// also: {type:"input_mouse",eventType:"mousePressed",x,y,button:"left",clickCount:1}
+```
+
+This is the difference between watching a slideshow and playing the game. Reserve
+screenshots for one-off checks; use the WS for any sustained real-time control.
+
 ## Waiting (read this)
 
 Agents fail more often from bad waits than from bad selectors. Pick the