browser-automation-cli 0.7.2__tar.gz → 0.8.0__tar.gz

{browser_automation_cli-0.7.2 → browser_automation_cli-0.8.0}/Cargo.lock

@@ -34,7 +34,7 @@ dependencies = [
 
 [[package]]
 name = "browser-cli"
-version = "0.7.2"
+version = "0.8.0"
 dependencies = [
  "base64",
  "futures-util",

{browser_automation_cli-0.7.2 → browser_automation_cli-0.8.0}/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "browser-cli"
-version = "0.7.2"
+version = "0.8.0"
 edition = "2021"
 description = "Browser automation daemon + CLI for coding agents. Persistent sessions, no MCP, no extensions."
 license = "MIT"

{browser_automation_cli-0.7.2 → browser_automation_cli-0.8.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: browser-automation-cli
-Version: 0.7.2
+Version: 0.8.0
 Classifier: Programming Language :: Rust
 Classifier: Programming Language :: Python :: 3
 Classifier: Operating System :: MacOS
@@ -141,7 +141,7 @@ All print JSON (snapshot prints text). Add `-s` / `--snapshot` to any action to
 | :-- | :-- |
 | `navigate <url> [--wait load\|domcontentloaded\|networkidle]` | Returns as soon as the page is usable; never fails on a slow `networkidle` |
 | `snapshot [scope] [--all] [--max N] [--json]` | Visible interactive elements + headings. `--all` adds text blocks, `--json` gives boxes and unique selectors |
-| `click <target> [--double]` | |
+| `click <target> [--double]` | Also `click --at X,Y` to click raw viewport pixels (canvas / vision — no DOM target; screenshot first, pixels map 1:1) |
 | `type <target> <text> [--sequential] [--submit]` | `fill()` by default; `--sequential` sends key events (autocomplete); `--submit` presses Enter |
 | `press <key> [target]` | `Enter`, `Tab`, `Control+a`, … |
 | `hover <target>` | |
@@ -155,7 +155,7 @@ All print JSON (snapshot prints text). Add `-s` / `--snapshot` to any action to
 | `back` / `forward` | |
 | `batch` | JSON lines on stdin, run in one round-trip, stop at first failure |
 
-**Targets:** `@e12` (ref from snapshot — preferred) · CSS selector · `text=Create` · `role=button[name=Create]` · `label=Email` · `placeholder=Search` · or flags `--text / --role [--name] / --label / --placeholder`. Ambiguous CSS selectors are refused (strict mode) instead of clicking the first match.
+**Targets:** `@e12` (ref from snapshot — preferred) · CSS selector · `text=Create` · `role=button[name=Create]` · `label=Email` · `placeholder=Search` · or flags `--text / --role [--name] / --label / --placeholder`. Ambiguous CSS selectors are refused (strict mode) instead of clicking the first match. For canvas / vision cases with no DOM target, `click --at X,Y` clicks raw viewport pixels (take a `screenshot` first; its pixels map 1:1 to click coordinates).
 
 ***
 

{browser_automation_cli-0.7.2 → browser_automation_cli-0.8.0}/README.md

@@ -126,7 +126,7 @@ All print JSON (snapshot prints text). Add `-s` / `--snapshot` to any action to
 | :-- | :-- |
 | `navigate <url> [--wait load\|domcontentloaded\|networkidle]` | Returns as soon as the page is usable; never fails on a slow `networkidle` |
 | `snapshot [scope] [--all] [--max N] [--json]` | Visible interactive elements + headings. `--all` adds text blocks, `--json` gives boxes and unique selectors |
-| `click <target> [--double]` | |
+| `click <target> [--double]` | Also `click --at X,Y` to click raw viewport pixels (canvas / vision — no DOM target; screenshot first, pixels map 1:1) |
 | `type <target> <text> [--sequential] [--submit]` | `fill()` by default; `--sequential` sends key events (autocomplete); `--submit` presses Enter |
 | `press <key> [target]` | `Enter`, `Tab`, `Control+a`, … |
 | `hover <target>` | |
@@ -140,7 +140,7 @@ All print JSON (snapshot prints text). Add `-s` / `--snapshot` to any action to
 | `back` / `forward` | |
 | `batch` | JSON lines on stdin, run in one round-trip, stop at first failure |
 
-**Targets:** `@e12` (ref from snapshot — preferred) · CSS selector · `text=Create` · `role=button[name=Create]` · `label=Email` · `placeholder=Search` · or flags `--text / --role [--name] / --label / --placeholder`. Ambiguous CSS selectors are refused (strict mode) instead of clicking the first match.
+**Targets:** `@e12` (ref from snapshot — preferred) · CSS selector · `text=Create` · `role=button[name=Create]` · `label=Email` · `placeholder=Search` · or flags `--text / --role [--name] / --label / --placeholder`. Ambiguous CSS selectors are refused (strict mode) instead of clicking the first match. For canvas / vision cases with no DOM target, `click --at X,Y` clicks raw viewport pixels (take a `screenshot` first; its pixels map 1:1 to click coordinates).
 
 ***
 

{browser_automation_cli-0.7.2 → browser_automation_cli-0.8.0}/SKILL.md

@@ -73,6 +73,7 @@ browser shutdown
 browser <id> navigate <url> [-s]        # -s on any action: append a fresh snapshot
 browser <id> snapshot [scope] [--all] [--max N] [--json]
 browser <id> click <target> [--double] [-s]
+browser <id> click --at X,Y [--double]   # click raw viewport pixels (canvas / vision — no DOM target)
 browser <id> type <target> <text> [--sequential] [--submit] [-s]   # alias: fill
 browser <id> press <key> [target]       # Enter, Tab, Escape, Control+a
 browser <id> hover <target>
@@ -87,7 +88,7 @@ browser <id> back | forward
 browser <id> batch                      # JSON lines on stdin, one round-trip
 ```
 
-**Targets:** `@e12` (ref from snapshot — preferred) · `--text "Create"` · `--role button --name Create` · `--label "Email"` · `--placeholder Search` · CSS selector. Ambiguous CSS selectors are refused; use a ref or text.
+**Targets:** `@e12` (ref from snapshot — preferred) · `--text "Create"` · `--role button --name Create` · `--label "Email"` · `--placeholder Search` · CSS selector. Ambiguous CSS selectors are refused; use a ref or text. For canvas / vision cases with no DOM target, `click --at X,Y` clicks raw viewport pixels (take a `screenshot` first; its pixels map 1:1 to click coordinates).
 
 ## Snapshot Format
 

{browser_automation_cli-0.7.2 → browser_automation_cli-0.8.0}/install.sh

@@ -1,7 +1,7 @@
 #!/bin/sh
 # Install the Rust browser/browser-daemon binaries from the GitHub release into ~/.local/bin.
 set -e
-V="${BROWSER_CLI_VERSION:-0.7.2}"; OS=$(uname -s | tr A-Z a-z); ARCH=$(uname -m)
+V="${BROWSER_CLI_VERSION:-0.8.0}"; OS=$(uname -s | tr A-Z a-z); ARCH=$(uname -m)
 URL="https://github.com/jshan9078/browser-automation-cli/releases/download/v$V/browser-cli-$V-$OS-$ARCH.tar.gz"
 DEST="${BROWSER_CLI_BIN:-$HOME/.local/bin}"; mkdir -p "$DEST"
 echo "Downloading $URL"; curl -fsSL "$URL" | tar -C "$DEST" -xzf -

{browser_automation_cli-0.7.2 → browser_automation_cli-0.8.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "maturin"
 
 [project]
 name = "browser-automation-cli"
-version = "0.7.2"
+version = "0.8.0"
 description = "Browser automation daemon + CLI for coding agents. Persistent sessions, no MCP, no extensions."
 readme = "README.md"
 requires-python = ">=3.9"

{browser_automation_cli-0.7.2 → browser_automation_cli-0.8.0}/src/actions.rs

@@ -189,8 +189,14 @@ pub async fn snapshot_p(page: &Page, p: &Map<String, Value>) -> Value {
 }
 
 pub async fn click(page: &Page, p: &Map<String, Value>) -> Value {
-    let r = match page.resolve(&target_of(p), ACTION_TIMEOUT_MS).await { Ok(r) => r, Err(e) => return fail(e) };
-    let (x, y) = (r["x"].as_f64().unwrap_or(0.0), r["y"].as_f64().unwrap_or(0.0));
+    // Raw viewport pixel click (`click --at X,Y`) when x/y are given; else resolve a DOM target.
+    let (x, y) = match (p.get("x").and_then(|v| v.as_f64()), p.get("y").and_then(|v| v.as_f64())) {
+        (Some(x), Some(y)) => (x, y),
+        _ => {
+            let r = match page.resolve(&target_of(p), ACTION_TIMEOUT_MS).await { Ok(r) => r, Err(e) => return fail(e) };
+            (r["x"].as_f64().unwrap_or(0.0), r["y"].as_f64().unwrap_or(0.0))
+        }
+    };
     if let Err(e) = page.click_at(x, y, if flag(p, "double") { 2 } else { 1 }).await { return fail(e); }
     page.after(ok(), flag(p, "snap"), true).await
 }

{browser_automation_cli-0.7.2 → browser_automation_cli-0.8.0}/src/main.rs

@@ -42,7 +42,7 @@ Page commands (all print JSON; add -s/--snapshot to include a fresh snapshot in
   browser <id> snapshot [scope-selector] [--all] [--max N] [--json]
                                          Interactive elements as "@e12 button "Create"" lines.
                                          --all adds text blocks; --json gives structured output
-  browser <id> click <target> [--double]
+  browser <id> click <target> [--double]      (or `--at X,Y` to click raw viewport pixels — canvas/vision)
   browser <id> type <target> <text> [--sequential] [--submit]   (alias: fill)
   browser <id> press <key> [target]      e.g. Enter, Tab, Control+a
   browser <id> hover <target>
@@ -128,7 +128,7 @@ fn parse_flags(args: &[String], bools: &[&str], valued: &[&str]) -> (Vec<String>
 }
 
 const BOOLS: &[&str] = &["-s", "--snapshot", "--all", "--json", "--double", "--sequential", "--submit", "--gone", "-f", "--full-page", "--clear", "--table"];
-const VALUED: &[&str] = &["--text", "--role", "--name", "--label", "--placeholder", "--wait", "--max", "--timeout", "-o", "--output", "-q", "--quality", "--selector", "--format"];
+const VALUED: &[&str] = &["--text", "--role", "--name", "--label", "--placeholder", "--wait", "--max", "--timeout", "-o", "--output", "-q", "--quality", "--selector", "--format", "--at"];
 
 fn target_params(f: &HashMap<String, Value>) -> Map<String, Value> {
     let mut p = Map::new();
@@ -166,7 +166,19 @@ fn build(sid: &str, action: &str, rest: &[String]) -> Option<Value> {
             req("snapshot", p)
         }
         "click" => {
-            p.insert("selector".into(), json!(sel(0)));
+            // `--at X,Y` clicks raw viewport pixels (for canvas / vision tasks with no DOM target);
+            // otherwise resolve a normal target (@ref / selector / --text / ...).
+            if let Some(at) = f.get("at").and_then(|v| v.as_str()) {
+                let mut it = at.split(',');
+                let x = it.next().and_then(|s| s.trim().parse::<f64>().ok());
+                let y = it.next().and_then(|s| s.trim().parse::<f64>().ok());
+                match (x, y) {
+                    (Some(x), Some(y)) => { p.insert("x".into(), json!(x)); p.insert("y".into(), json!(y)); }
+                    _ => return None,
+                }
+            } else {
+                p.insert("selector".into(), json!(sel(0)));
+            }
             if f.contains_key("double") { p.insert("double".into(), json!(true)); }
             req("click", p)
         }