Merge main into ctate/custom-headers

support serverless environments (#29 )
* add --executable-path * tests * test vercel * fixes
2026-01-12 11:57:43 -06:00 · 2026-01-12 11:41:25 -06:00 · 2026-01-12 11:40:52 -06:00 · 2026-01-12 11:26:45 -06:00 · 2026-01-12 11:22:46 -06:00 · 2026-01-12 10:52:31 -06:00
18 changed files with 1768 additions and 36 deletions
@@ -83,3 +83,106 @@ jobs:

      - name: Build release binary
        run: cargo build --release --manifest-path cli/Cargo.toml --target ${{ matrix.target }}
+
+      - name: Run Rust tests
+        run: cargo test --manifest-path cli/Cargo.toml --target ${{ matrix.target }}
+
+  windows-integration:
+    name: Windows Integration Test
+    runs-on: windows-latest
+    needs: rust
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Setup pnpm
+        uses: pnpm/action-setup@v4
+        with:
+          version: 9
+
+      - name: Setup Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: 22
+          cache: pnpm
+
+      - name: Setup Rust toolchain
+        uses: dtolnay/rust-toolchain@stable
+        with:
+          targets: x86_64-pc-windows-msvc
+
+      - name: Cache Cargo dependencies
+        uses: actions/cache@v4
+        with:
+          path: |
+            ~/.cargo/bin/
+            ~/.cargo/registry/index/
+            ~/.cargo/registry/cache/
+            ~/.cargo/git/db/
+            cli/target/
+          key: windows-cargo-x86_64-pc-windows-msvc-${{ hashFiles('cli/Cargo.lock') }}
+          restore-keys: |
+            windows-cargo-x86_64-pc-windows-msvc-
+
+      - name: Build Rust CLI
+        run: cargo build --release --manifest-path cli/Cargo.toml --target x86_64-pc-windows-msvc
+
+      - name: Install npm dependencies
+        run: pnpm install
+
+      - name: Build TypeScript
+        run: pnpm build
+
+      - name: Copy CLI binary to bin directory
+        run: |
+          Copy-Item cli/target/x86_64-pc-windows-msvc/release/agent-browser.exe bin/agent-browser-win32-x64.exe
+
+      - name: Test agent-browser install command
+        run: |
+          $env:PATH = "$pwd\bin;$env:PATH"
+          bin/agent-browser-win32-x64.exe install
+        shell: pwsh
+
+      - name: Verify Chromium was installed
+        run: |
+          $playwrightPath = "$env:LOCALAPPDATA\ms-playwright"
+          if (Test-Path $playwrightPath) {
+            Write-Host "Playwright browsers installed at: $playwrightPath"
+            Get-ChildItem $playwrightPath -Recurse -Depth 2 | Select-Object -First 20
+          } else {
+            Write-Error "Playwright browsers not found!"
+            exit 1
+          }
+        shell: pwsh
+
+  serverless-chromium:
+    name: Serverless Chromium (@sparticuz/chromium)
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Setup pnpm
+        uses: pnpm/action-setup@v4
+        with:
+          version: 9
+
+      - name: Setup Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: 22
+          cache: pnpm
+
+      - name: Install dependencies
+        run: pnpm install
+
+      - name: Install @sparticuz/chromium
+        run: pnpm add -D @sparticuz/chromium
+
+      - name: Build TypeScript
+        run: pnpm build
+
+      - name: Run serverless integration test
+        run: pnpm exec vitest run test/serverless.test.ts
@@ -55,13 +55,13 @@ agent-browser find role button click --name "Submit"
 ### Core Commands

 ```bash
-agent-browser open <url>              # Navigate to URL
+agent-browser open <url>              # Navigate to URL (aliases: goto, navigate)
 agent-browser click <sel>             # Click element
 agent-browser dblclick <sel>          # Double-click element
 agent-browser focus <sel>             # Focus element
 agent-browser type <sel> <text>       # Type into element
 agent-browser fill <sel> <text>       # Clear and fill
-agent-browser press <key>             # Press key (Enter, Tab, Control+a)
+agent-browser press <key>             # Press key (Enter, Tab, Control+a) (alias: key)
 agent-browser keydown <key>           # Hold key down
 agent-browser keyup <key>             # Release key
 agent-browser hover <sel>             # Hover element
@@ -69,14 +69,14 @@ agent-browser select <sel> <val>      # Select dropdown option
 agent-browser check <sel>             # Check checkbox
 agent-browser uncheck <sel>           # Uncheck checkbox
 agent-browser scroll <dir> [px]       # Scroll (up/down/left/right)
-agent-browser scrollintoview <sel>    # Scroll element into view
+agent-browser scrollintoview <sel>    # Scroll element into view (alias: scrollinto)
 agent-browser drag <src> <tgt>        # Drag and drop
 agent-browser upload <sel> <files>    # Upload files
 agent-browser screenshot [path]       # Take screenshot (--full for full page)
 agent-browser pdf <path>              # Save as PDF
 agent-browser snapshot                # Accessibility tree with refs (best for AI)
 agent-browser eval <js>               # Run JavaScript
-agent-browser close                   # Close browser
+agent-browser close                   # Close browser (aliases: quit, exit)
 ```

 ### Get Info
@@ -129,9 +129,9 @@ agent-browser find nth 2 "a" text
 ### Wait

 ```bash
-agent-browser wait <selector>         # Wait for element
-agent-browser wait <ms>               # Wait for time
-agent-browser wait --text "Welcome"   # Wait for text
+agent-browser wait <selector>         # Wait for element to be visible
+agent-browser wait <ms>               # Wait for time (milliseconds)
+agent-browser wait --text "Welcome"   # Wait for text to appear
 agent-browser wait --url "**/dash"    # Wait for URL pattern
 agent-browser wait --load networkidle # Wait for load state
 agent-browser wait --fn "window.ready === true"  # Wait for JS condition
@@ -253,6 +253,10 @@ AGENT_BROWSER_SESSION=agent1 agent-browser click "#btn"

 # List active sessions
 agent-browser session list
+# Output:
+# Active sessions:
+# -> default
+#    agent1

 # Show current session
 agent-browser session
@@ -289,6 +293,8 @@ agent-browser snapshot -i -c -d 5         # Combine options
 | Option | Description |
 |--------|-------------|
 | `--session <name>` | Use isolated session (or `AGENT_BROWSER_SESSION` env) |
+| `--headers <json>` | Set HTTP headers scoped to the URL's origin |
+| `--executable-path <path>` | Custom browser executable (or `AGENT_BROWSER_EXECUTABLE_PATH` env) |
 | `--json` | JSON output (for agents) |
 | `--full, -f` | Full page screenshot |
 | `--name, -n` | Locator name filter |
@@ -383,6 +389,74 @@ agent-browser open example.com --headed

 This opens a visible browser window instead of running headless.

+## Authenticated Sessions
+
+Use `--headers` to set HTTP headers for a specific origin, enabling authentication without login flows:
+
+```bash
+# Headers are scoped to api.example.com only
+agent-browser open api.example.com --headers '{"Authorization": "Bearer <token>"}'
+
+# Requests to api.example.com include the auth header
+agent-browser snapshot -i --json
+agent-browser click @e2
+
+# Navigate to another domain - headers are NOT sent (safe!)
+agent-browser open other-site.com
+```
+
+This is useful for:
+- **Skipping login flows** - Authenticate via headers instead of UI
+- **Switching users** - Start new sessions with different auth tokens
+- **API testing** - Access protected endpoints directly
+- **Security** - Headers are scoped to the origin, not leaked to other domains
+
+To set headers for multiple origins, use `--headers` with each `open` command:
+
+```bash
+agent-browser open api.example.com --headers '{"Authorization": "Bearer token1"}'
+agent-browser open api.acme.com --headers '{"Authorization": "Bearer token2"}'
+```
+
+For global headers (all domains), use `set headers`:
+
+```bash
+agent-browser set headers '{"X-Custom-Header": "value"}'
+```
+
+## Custom Browser Executable
+
+Use a custom browser executable instead of the bundled Chromium. This is useful for:
+- **Serverless deployment**: Use lightweight Chromium builds like `@sparticuz/chromium` (~50MB vs ~684MB)
+- **System browsers**: Use an existing Chrome/Chromium installation
+- **Custom builds**: Use modified browser builds
+
+### CLI Usage
+
+```bash
+# Via flag
+agent-browser --executable-path /path/to/chromium open example.com
+
+# Via environment variable
+AGENT_BROWSER_EXECUTABLE_PATH=/path/to/chromium agent-browser open example.com
+```
+
+### Serverless Example (Vercel/AWS Lambda)
+
+```typescript
+import chromium from '@sparticuz/chromium';
+import { BrowserManager } from 'agent-browser';
+
+export async function handler() {
+  const browser = new BrowserManager();
+  await browser.launch({
+    executablePath: await chromium.executablePath(),
+    headless: true,
+  });
+  // ... use browser
+}
+```
+
 ## Architecture

 agent-browser uses a client-daemon architecture:
@@ -393,15 +467,17 @@ agent-browser uses a client-daemon architecture:

 The daemon starts automatically on first command and persists between commands for fast subsequent operations.

+**Browser Engine:** Uses Chromium by default. The daemon also supports Firefox and WebKit via the Playwright protocol.
+
 ## Platforms

 | Platform | Binary | Fallback |
 |----------|--------|----------|
-| macOS ARM64 | ✅ Native Rust | Node.js |
-| macOS x64 | ✅ Native Rust | Node.js |
-| Linux ARM64 | ✅ Native Rust | Node.js |
-| Linux x64 | ✅ Native Rust | Node.js |
-| Windows | - | Node.js |
+| macOS ARM64 | Native Rust | Node.js |
+| macOS x64 | Native Rust | Node.js |
+| Linux ARM64 | Native Rust | Node.js |
+| Linux x64 | Native Rust | Node.js |
+| Windows x64 | Native Rust | Node.js |

 ## Usage with AI Agents

@@ -4,7 +4,7 @@ version = 4

 [[package]]
 name = "agent-browser"
-version = "0.4.2"
+version = "0.4.3"
 dependencies = [
 "libc",
 "serde",
@@ -1,6 +1,6 @@
 [package]
 name = "agent-browser"
-version = "0.4.2"
+version = "0.4.3"
 edition = "2021"
 description = "Fast browser automation CLI for AI agents"
 license = "Apache-2.0"
@@ -80,7 +80,14 @@ pub fn parse_command(args: &[String], flags: &Flags) -> Result<Value, ParseError
            } else {
                format!("https://{}", url)
            };
-            Ok(json!({ "id": id, "action": "navigate", "url": url }))
+            let mut nav_cmd = json!({ "id": id, "action": "navigate", "url": url });
+            // If --headers flag is set, include headers (scoped to this origin)
+            if let Some(ref headers_json) = flags.headers {
+                if let Ok(headers) = serde_json::from_str::<serde_json::Value>(headers_json) {
+                    nav_cmd["headers"] = headers;
+                }
+            }
+            Ok(nav_cmd)
        }
        "back" => Ok(json!({ "id": id, "action": "back" })),
        "forward" => Ok(json!({ "id": id, "action": "forward" })),
@@ -212,6 +219,44 @@ pub fn parse_command(args: &[String], flags: &Flags) -> Result<Value, ParseError

        // === Wait ===
        "wait" => {
+            // Check for --url flag: wait --url "**/dashboard"
+            if let Some(idx) = rest.iter().position(|&s| s == "--url" || s == "-u") {
+                let url = rest.get(idx + 1).ok_or_else(|| ParseError::MissingArguments {
+                    context: "wait --url".to_string(),
+                    usage: "wait --url <pattern>",
+                })?;
+                return Ok(json!({ "id": id, "action": "waitforurl", "url": url }));
+            }
+            
+            // Check for --load flag: wait --load networkidle
+            if let Some(idx) = rest.iter().position(|&s| s == "--load" || s == "-l") {
+                let state = rest.get(idx + 1).ok_or_else(|| ParseError::MissingArguments {
+                    context: "wait --load".to_string(),
+                    usage: "wait --load <state>",
+                })?;
+                return Ok(json!({ "id": id, "action": "waitforloadstate", "state": state }));
+            }
+            
+            // Check for --fn flag: wait --fn "window.ready === true"
+            if let Some(idx) = rest.iter().position(|&s| s == "--fn" || s == "-f") {
+                let expr = rest.get(idx + 1).ok_or_else(|| ParseError::MissingArguments {
+                    context: "wait --fn".to_string(),
+                    usage: "wait --fn <expression>",
+                })?;
+                return Ok(json!({ "id": id, "action": "waitforfunction", "expression": expr }));
+            }
+            
+            // Check for --text flag: wait --text "Welcome"
+            if let Some(idx) = rest.iter().position(|&s| s == "--text" || s == "-t") {
+                let text = rest.get(idx + 1).ok_or_else(|| ParseError::MissingArguments {
+                    context: "wait --text".to_string(),
+                    usage: "wait --text <text>",
+                })?;
+                // Use getByText locator to wait for text to appear
+                return Ok(json!({ "id": id, "action": "wait", "selector": format!("text={}", text) }));
+            }
+            
+            // Default: selector or timeout
            if let Some(arg) = rest.get(0) {
                if arg.parse::<u64>().is_ok() {
                    Ok(json!({ "id": id, "action": "wait", "timeout": arg.parse::<u64>().unwrap() }))
@@ -221,7 +266,7 @@ pub fn parse_command(args: &[String], flags: &Flags) -> Result<Value, ParseError
            } else {
                Err(ParseError::MissingArguments {
                    context: "wait".to_string(),
-                    usage: "wait <selector|ms>",
+                    usage: "wait <selector|ms|--url|--load|--fn|--text>",
                })
            }
        }
@@ -728,7 +773,13 @@ fn parse_set(rest: &[&str], id: &str) -> Result<Value, ParseError> {
                context: "set headers".to_string(),
                usage: "set headers <json>",
            })?;
-            Ok(json!({ "id": id, "action": "headers", "headers": headers_json }))
+            // Parse the JSON string into an object
+            let headers: serde_json::Value = serde_json::from_str(headers_json)
+                .map_err(|_| ParseError::MissingArguments {
+                    context: "set headers".to_string(),
+                    usage: "set headers <json> (must be valid JSON object)",
+                })?;
+            Ok(json!({ "id": id, "action": "headers", "headers": headers }))
        }
        Some("credentials") | Some("auth") => {
            let user = rest.get(1).ok_or_else(|| ParseError::MissingArguments {
@@ -848,6 +899,8 @@ mod tests {
            full: false,
            headed: false,
            debug: false,
+            headers: None,
+            executable_path: None,
        }
    }

@@ -974,6 +1027,81 @@ mod tests {
        assert_eq!(cmd["url"], "https://example.com");
    }

+    #[test]
+    fn test_navigate_with_headers() {
+        let mut flags = default_flags();
+        flags.headers = Some(r#"{"Authorization": "Bearer token"}"#.to_string());
+        let cmd = parse_command(&args("open api.example.com"), &flags).unwrap();
+        assert_eq!(cmd["action"], "navigate");
+        assert_eq!(cmd["url"], "https://api.example.com");
+        assert_eq!(cmd["headers"]["Authorization"], "Bearer token");
+    }
+
+    #[test]
+    fn test_navigate_with_multiple_headers() {
+        let mut flags = default_flags();
+        flags.headers = Some(r#"{"Authorization": "Bearer token", "X-Custom": "value"}"#.to_string());
+        let cmd = parse_command(&args("open api.example.com"), &flags).unwrap();
+        assert_eq!(cmd["headers"]["Authorization"], "Bearer token");
+        assert_eq!(cmd["headers"]["X-Custom"], "value");
+    }
+
+    #[test]
+    fn test_navigate_without_headers_flag() {
+        let cmd = parse_command(&args("open example.com"), &default_flags()).unwrap();
+        assert_eq!(cmd["action"], "navigate");
+        // headers should not be present when flag is not set
+        assert!(cmd.get("headers").is_none());
+    }
+
+    #[test]
+    fn test_navigate_with_invalid_headers_json() {
+        let mut flags = default_flags();
+        flags.headers = Some("not valid json".to_string());
+        let cmd = parse_command(&args("open api.example.com"), &flags).unwrap();
+        // Invalid JSON should result in no headers field (graceful handling)
+        assert!(cmd.get("headers").is_none());
+    }
+
+    // === Set Headers Tests ===
+
+    #[test]
+    fn test_set_headers_parses_json() {
+        let input: Vec<String> = vec![
+            "set".to_string(),
+            "headers".to_string(),
+            r#"{"Authorization":"Bearer token"}"#.to_string(),
+        ];
+        let cmd = parse_command(&input, &default_flags()).unwrap();
+        assert_eq!(cmd["action"], "headers");
+        // Headers should be an object, not a string
+        assert!(cmd["headers"].is_object());
+        assert_eq!(cmd["headers"]["Authorization"], "Bearer token");
+    }
+
+    #[test]
+    fn test_set_headers_with_multiple_values() {
+        let input: Vec<String> = vec![
+            "set".to_string(),
+            "headers".to_string(),
+            r#"{"Authorization": "Bearer token", "X-Custom": "value"}"#.to_string(),
+        ];
+        let cmd = parse_command(&input, &default_flags()).unwrap();
+        assert_eq!(cmd["headers"]["Authorization"], "Bearer token");
+        assert_eq!(cmd["headers"]["X-Custom"], "value");
+    }
+
+    #[test]
+    fn test_set_headers_invalid_json_error() {
+        let input: Vec<String> = vec![
+            "set".to_string(),
+            "headers".to_string(),
+            "not-valid-json".to_string(),
+        ];
+        let result = parse_command(&input, &default_flags());
+        assert!(result.is_err());
+    }
+
    #[test]
    fn test_back() {
        let cmd = parse_command(&args("back"), &default_flags()).unwrap();
@@ -1090,6 +1218,57 @@ mod tests {
        assert_eq!(cmd["maxDepth"], 3);
    }

+    // === Wait ===
+
+    #[test]
+    fn test_wait_selector() {
+        let cmd = parse_command(&args("wait #element"), &default_flags()).unwrap();
+        assert_eq!(cmd["action"], "wait");
+        assert_eq!(cmd["selector"], "#element");
+    }
+
+    #[test]
+    fn test_wait_timeout() {
+        let cmd = parse_command(&args("wait 5000"), &default_flags()).unwrap();
+        assert_eq!(cmd["action"], "wait");
+        assert_eq!(cmd["timeout"], 5000);
+    }
+
+    #[test]
+    fn test_wait_url() {
+        let cmd = parse_command(&args("wait --url **/dashboard"), &default_flags()).unwrap();
+        assert_eq!(cmd["action"], "waitforurl");
+        assert_eq!(cmd["url"], "**/dashboard");
+    }
+
+    #[test]
+    fn test_wait_load() {
+        let cmd = parse_command(&args("wait --load networkidle"), &default_flags()).unwrap();
+        assert_eq!(cmd["action"], "waitforloadstate");
+        assert_eq!(cmd["state"], "networkidle");
+    }
+
+    #[test]
+    fn test_wait_load_missing_state() {
+        let result = parse_command(&args("wait --load"), &default_flags());
+        assert!(result.is_err());
+        assert!(matches!(result.unwrap_err(), ParseError::MissingArguments { .. }));
+    }
+
+    #[test]
+    fn test_wait_fn() {
+        let cmd = parse_command(&args("wait --fn window.ready"), &default_flags()).unwrap();
+        assert_eq!(cmd["action"], "waitforfunction");
+        assert_eq!(cmd["expression"], "window.ready");
+    }
+
+    #[test]
+    fn test_wait_text() {
+        let cmd = parse_command(&args("wait --text Welcome"), &default_flags()).unwrap();
+        assert_eq!(cmd["action"], "wait");
+        assert_eq!(cmd["selector"], "text=Welcome");
+    }
+
    // === Unknown command ===

    #[test]
@@ -153,9 +153,15 @@ fn daemon_ready(session: &str) -> bool {
    }
 }

-pub fn ensure_daemon(session: &str, headed: bool) -> Result<(), String> {
+/// Result of ensure_daemon indicating whether a new daemon was started
+pub struct DaemonResult {
+    /// True if we connected to an existing daemon, false if we started a new one
+    pub already_running: bool,
+}
+
+pub fn ensure_daemon(session: &str, headed: bool, executable_path: Option<&str>) -> Result<DaemonResult, String> {
    if is_daemon_running(session) && daemon_ready(session) {
-        return Ok(());
+        return Ok(DaemonResult { already_running: true });
    }

    let exe_path = env::current_exe().map_err(|e| e.to_string())?;
@@ -186,6 +192,10 @@ pub fn ensure_daemon(session: &str, headed: bool) -> Result<(), String> {
            cmd.env("AGENT_BROWSER_HEADED", "1");
        }

+        if let Some(path) = executable_path {
+            cmd.env("AGENT_BROWSER_EXECUTABLE_PATH", path);
+        }
+
        // Create new process group and session to fully detach
        unsafe {
            cmd.pre_exec(|| {
@@ -206,8 +216,13 @@ pub fn ensure_daemon(session: &str, headed: bool) -> Result<(), String> {
    {
        use std::os::windows::process::CommandExt;
        
-        let mut cmd = Command::new("node");
-        cmd.arg(daemon_path)
+        // On Windows, use cmd.exe to run node to ensure proper PATH resolution.
+        // This handles cases where node.exe isn't directly in PATH but node.cmd is.
+        // Pass the entire command as a single string to /c to handle paths with spaces.
+        let cmd_string = format!("node \"{}\"", daemon_path.display());
+        let mut cmd = Command::new("cmd");
+        cmd.arg("/c")
+            .arg(&cmd_string)
            .env("AGENT_BROWSER_DAEMON", "1")
            .env("AGENT_BROWSER_SESSION", session);

@@ -215,6 +230,10 @@ pub fn ensure_daemon(session: &str, headed: bool) -> Result<(), String> {
            cmd.env("AGENT_BROWSER_HEADED", "1");
        }

+        if let Some(path) = executable_path {
+            cmd.env("AGENT_BROWSER_EXECUTABLE_PATH", path);
+        }
+
        // CREATE_NEW_PROCESS_GROUP | DETACHED_PROCESS
        const CREATE_NEW_PROCESS_GROUP: u32 = 0x00000200;
        const DETACHED_PROCESS: u32 = 0x00000008;
@@ -229,7 +248,7 @@ pub fn ensure_daemon(session: &str, headed: bool) -> Result<(), String> {

    for _ in 0..50 {
        if daemon_ready(session) {
-            return Ok(());
+            return Ok(DaemonResult { already_running: false });
        }
        thread::sleep(Duration::from_millis(100));
    }
@@ -6,6 +6,8 @@ pub struct Flags {
    pub headed: bool,
    pub debug: bool,
    pub session: String,
+    pub headers: Option<String>,
+    pub executable_path: Option<String>,
 }

 pub fn parse_flags(args: &[String]) -> Flags {
@@ -15,6 +17,8 @@ pub fn parse_flags(args: &[String]) -> Flags {
        headed: false,
        debug: false,
        session: env::var("AGENT_BROWSER_SESSION").unwrap_or_else(|_| "default".to_string()),
+        headers: None,
+        executable_path: env::var("AGENT_BROWSER_EXECUTABLE_PATH").ok(),
    };

    let mut i = 0;
@@ -30,6 +34,18 @@ pub fn parse_flags(args: &[String]) -> Flags {
                    i += 1;
                }
            }
+            "--headers" => {
+                if let Some(h) = args.get(i + 1) {
+                    flags.headers = Some(h.clone());
+                    i += 1;
+                }
+            }
+            "--executable-path" => {
+                if let Some(s) = args.get(i + 1) {
+                    flags.executable_path = Some(s.clone());
+                    i += 1;
+                }
+            }
            _ => {}
        }
        i += 1;
@@ -43,13 +59,15 @@ pub fn clean_args(args: &[String]) -> Vec<String> {

    // Global flags that should be stripped from command args
    const GLOBAL_FLAGS: &[&str] = &["--json", "--full", "--headed", "--debug"];
+    // Global flags that take a value (need to skip the next arg too)
+    const GLOBAL_FLAGS_WITH_VALUE: &[&str] = &["--session", "--headers", "--executable-path"];

    for arg in args.iter() {
        if skip_next {
            skip_next = false;
            continue;
        }
-        if arg == "--session" {
+        if GLOBAL_FLAGS_WITH_VALUE.contains(&arg.as_str()) {
            skip_next = true;
            continue;
        }
@@ -61,3 +79,111 @@ pub fn clean_args(args: &[String]) -> Vec<String> {
    }
    result
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    fn args(s: &str) -> Vec<String> {
+        s.split_whitespace().map(String::from).collect()
+    }
+
+    #[test]
+    fn test_parse_headers_flag() {
+        let flags = parse_flags(&args(r#"open example.com --headers {"Auth":"token"}"#));
+        assert_eq!(flags.headers, Some(r#"{"Auth":"token"}"#.to_string()));
+    }
+
+    #[test]
+    fn test_parse_headers_flag_with_spaces() {
+        // Headers JSON is passed as a single quoted argument in shell
+        let input: Vec<String> = vec![
+            "open".to_string(),
+            "example.com".to_string(),
+            "--headers".to_string(),
+            r#"{"Authorization": "Bearer token"}"#.to_string(),
+        ];
+        let flags = parse_flags(&input);
+        assert_eq!(flags.headers, Some(r#"{"Authorization": "Bearer token"}"#.to_string()));
+    }
+
+    #[test]
+    fn test_parse_no_headers_flag() {
+        let flags = parse_flags(&args("open example.com"));
+        assert!(flags.headers.is_none());
+    }
+
+    #[test]
+    fn test_clean_args_removes_headers() {
+        let input: Vec<String> = vec![
+            "open".to_string(),
+            "example.com".to_string(),
+            "--headers".to_string(),
+            r#"{"Auth":"token"}"#.to_string(),
+        ];
+        let clean = clean_args(&input);
+        assert_eq!(clean, vec!["open", "example.com"]);
+    }
+
+    #[test]
+    fn test_clean_args_removes_headers_at_start() {
+        let input: Vec<String> = vec![
+            "--headers".to_string(),
+            r#"{"Auth":"token"}"#.to_string(),
+            "open".to_string(),
+            "example.com".to_string(),
+        ];
+        let clean = clean_args(&input);
+        assert_eq!(clean, vec!["open", "example.com"]);
+    }
+
+    #[test]
+    fn test_headers_with_other_flags() {
+        let input: Vec<String> = vec![
+            "open".to_string(),
+            "example.com".to_string(),
+            "--headers".to_string(),
+            r#"{"Auth":"token"}"#.to_string(),
+            "--json".to_string(),
+            "--headed".to_string(),
+        ];
+        let flags = parse_flags(&input);
+        assert_eq!(flags.headers, Some(r#"{"Auth":"token"}"#.to_string()));
+        assert!(flags.json);
+        assert!(flags.headed);
+        
+        let clean = clean_args(&input);
+        assert_eq!(clean, vec!["open", "example.com"]);
+    }
+
+    #[test]
+    fn test_parse_executable_path_flag() {
+        let flags = parse_flags(&args("--executable-path /path/to/chromium open example.com"));
+        assert_eq!(flags.executable_path, Some("/path/to/chromium".to_string()));
+    }
+
+    #[test]
+    fn test_parse_executable_path_flag_no_value() {
+        let flags = parse_flags(&args("--executable-path"));
+        assert_eq!(flags.executable_path, None);
+    }
+
+    #[test]
+    fn test_clean_args_removes_executable_path() {
+        let cleaned = clean_args(&args("--executable-path /path/to/chromium open example.com"));
+        assert_eq!(cleaned, vec!["open", "example.com"]);
+    }
+
+    #[test]
+    fn test_clean_args_removes_executable_path_with_other_flags() {
+        let cleaned = clean_args(&args("--json --executable-path /path/to/chromium --headed open example.com"));
+        assert_eq!(cleaned, vec!["open", "example.com"]);
+    }
+
+    #[test]
+    fn test_parse_flags_with_session_and_executable_path() {
+        let flags = parse_flags(&args("--session test --executable-path /custom/chrome open example.com"));
+        assert_eq!(flags.session, "test");
+        assert_eq!(flags.executable_path, Some("/custom/chrome".to_string()));
+    }
+}
@@ -128,6 +128,16 @@ pub fn run_install(with_deps: bool) {
    }

    println!("\x1b[36mInstalling Chromium browser...\x1b[0m");
+    
+    // On Windows, we need to use cmd.exe to run npx because npx is actually npx.cmd
+    // and Command::new() doesn't resolve .cmd files the way the shell does.
+    // Pass the entire command as a single string to /c to handle paths with spaces.
+    #[cfg(windows)]
+    let status = Command::new("cmd")
+        .args(["/c", "npx playwright install chromium"])
+        .status();
+    
+    #[cfg(not(windows))]
    let status = Command::new("npx")
        .args(["playwright", "install", "chromium"])
        .status();
@@ -21,7 +21,7 @@ use commands::{gen_id, parse_command, ParseError};
 use connection::{ensure_daemon, send_command};
 use flags::{clean_args, parse_flags};
 use install::run_install;
-use output::{print_help, print_response};
+use output::{print_command_help, print_help, print_response};

 fn run_session(args: &[String], session: &str, json_mode: bool) {
    let subcommand = args.get(1).map(|s| s.as_str());
@@ -98,7 +98,19 @@ fn main() {
    let flags = parse_flags(&args);
    let clean = clean_args(&args);

-    if clean.is_empty() || args.iter().any(|a| a == "--help" || a == "-h") {
+    let has_help = args.iter().any(|a| a == "--help" || a == "-h");
+
+    if clean.is_empty() {
+        print_help();
+        return;
+    }
+
+    if has_help {
+        if let Some(cmd) = clean.get(0) {
+            if print_command_help(cmd) {
+                return;
+            }
+        }
        print_help();
        return;
    }
@@ -137,13 +149,23 @@ fn main() {
        }
    };

-    if let Err(e) = ensure_daemon(&flags.session, flags.headed) {
-        if flags.json {
-            println!(r#"{{"success":false,"error":"{}"}}"#, e);
-        } else {
-            eprintln!("\x1b[31m✗\x1b[0m {}", e);
+    let daemon_result = match ensure_daemon(&flags.session, flags.headed, flags.executable_path.as_deref()) {
+        Ok(result) => result,
+        Err(e) => {
+            if flags.json {
+                println!(r#"{{"success":false,"error":"{}"}}"#, e);
+            } else {
+                eprintln!("\x1b[31m✗\x1b[0m {}", e);
+            }
+            exit(1);
+        }
+    };
+
+    // Warn if executable_path was specified but daemon was already running
+    if daemon_result.already_running && flags.executable_path.is_some() {
+        if !flags.json {
+            eprintln!("\x1b[33m⚠\x1b[0m --executable-path ignored: daemon already running. Use 'agent-browser close' first to restart with new path.");
        }
-        exit(1);
    }

    // If --headed flag is set, send launch command first to switch to headed mode
@@ -145,6 +145,964 @@ pub fn print_response(resp: &Response, json_mode: bool) {
    }
 }

+/// Print command-specific help. Returns true if help was printed, false if command unknown.
+pub fn print_command_help(command: &str) -> bool {
+    let help = match command {
+        // === Navigation ===
+        "open" | "goto" | "navigate" => r##"
+agent-browser open - Navigate to a URL
+
+Usage: agent-browser open <url>
+
+Navigates the browser to the specified URL. If no protocol is provided,
+https:// is automatically prepended.
+
+Aliases: goto, navigate
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+  --headers <json>     Set HTTP headers (scoped to this origin)
+  --headed             Show browser window
+
+Examples:
+  agent-browser open example.com
+  agent-browser open https://github.com
+  agent-browser open localhost:3000
+  agent-browser open api.example.com --headers '{"Authorization": "Bearer token"}'
+    # ^ Headers only sent to api.example.com, not other domains
+"##,
+        "back" => r##"
+agent-browser back - Navigate back in history
+
+Usage: agent-browser back
+
+Goes back one page in the browser history, equivalent to clicking
+the browser's back button.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser back
+"##,
+        "forward" => r##"
+agent-browser forward - Navigate forward in history
+
+Usage: agent-browser forward
+
+Goes forward one page in the browser history, equivalent to clicking
+the browser's forward button.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser forward
+"##,
+        "reload" => r##"
+agent-browser reload - Reload the current page
+
+Usage: agent-browser reload
+
+Reloads the current page, equivalent to pressing F5 or clicking
+the browser's reload button.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser reload
+"##,
+
+        // === Core Actions ===
+        "click" => r##"
+agent-browser click - Click an element
+
+Usage: agent-browser click <selector>
+
+Clicks on the specified element. The selector can be a CSS selector,
+XPath, or an element reference from snapshot (e.g., @e1).
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser click "#submit-button"
+  agent-browser click @e1
+  agent-browser click "button.primary"
+  agent-browser click "//button[@type='submit']"
+"##,
+        "dblclick" => r##"
+agent-browser dblclick - Double-click an element
+
+Usage: agent-browser dblclick <selector>
+
+Double-clicks on the specified element. Useful for text selection
+or triggering double-click handlers.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser dblclick "#editable-text"
+  agent-browser dblclick @e5
+"##,
+        "fill" => r##"
+agent-browser fill - Clear and fill an input field
+
+Usage: agent-browser fill <selector> <text>
+
+Clears the input field and fills it with the specified text.
+This replaces any existing content in the field.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser fill "#email" "user@example.com"
+  agent-browser fill @e3 "Hello World"
+  agent-browser fill "input[name='search']" "query"
+"##,
+        "type" => r##"
+agent-browser type - Type text into an element
+
+Usage: agent-browser type <selector> <text>
+
+Types text into the specified element character by character.
+Unlike fill, this does not clear existing content first.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser type "#search" "hello"
+  agent-browser type @e2 "additional text"
+"##,
+        "hover" => r##"
+agent-browser hover - Hover over an element
+
+Usage: agent-browser hover <selector>
+
+Moves the mouse to hover over the specified element. Useful for
+triggering hover states or dropdown menus.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser hover "#dropdown-trigger"
+  agent-browser hover @e4
+"##,
+        "focus" => r##"
+agent-browser focus - Focus an element
+
+Usage: agent-browser focus <selector>
+
+Sets keyboard focus to the specified element.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser focus "#input-field"
+  agent-browser focus @e2
+"##,
+        "check" => r##"
+agent-browser check - Check a checkbox
+
+Usage: agent-browser check <selector>
+
+Checks a checkbox element. If already checked, no action is taken.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser check "#terms-checkbox"
+  agent-browser check @e7
+"##,
+        "uncheck" => r##"
+agent-browser uncheck - Uncheck a checkbox
+
+Usage: agent-browser uncheck <selector>
+
+Unchecks a checkbox element. If already unchecked, no action is taken.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser uncheck "#newsletter-opt-in"
+  agent-browser uncheck @e8
+"##,
+        "select" => r##"
+agent-browser select - Select a dropdown option
+
+Usage: agent-browser select <selector> <value>
+
+Selects an option in a <select> dropdown by its value attribute.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser select "#country" "US"
+  agent-browser select @e5 "option2"
+"##,
+        "drag" => r##"
+agent-browser drag - Drag and drop
+
+Usage: agent-browser drag <source> <target>
+
+Drags an element from source to target location.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser drag "#draggable" "#drop-zone"
+  agent-browser drag @e1 @e2
+"##,
+        "upload" => r##"
+agent-browser upload - Upload files
+
+Usage: agent-browser upload <selector> <files...>
+
+Uploads one or more files to a file input element.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser upload "#file-input" ./document.pdf
+  agent-browser upload @e3 ./image1.png ./image2.png
+"##,
+
+        // === Keyboard ===
+        "press" | "key" => r##"
+agent-browser press - Press a key or key combination
+
+Usage: agent-browser press <key>
+
+Presses a key or key combination. Supports special keys and modifiers.
+
+Aliases: key
+
+Special Keys:
+  Enter, Tab, Escape, Backspace, Delete, Space
+  ArrowUp, ArrowDown, ArrowLeft, ArrowRight
+  Home, End, PageUp, PageDown
+  F1-F12
+
+Modifiers (combine with +):
+  Control, Alt, Shift, Meta
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser press Enter
+  agent-browser press Tab
+  agent-browser press Control+a
+  agent-browser press Control+Shift+s
+  agent-browser press Escape
+"##,
+        "keydown" => r##"
+agent-browser keydown - Press a key down (without release)
+
+Usage: agent-browser keydown <key>
+
+Presses a key down without releasing it. Use keyup to release.
+Useful for holding modifier keys.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser keydown Shift
+  agent-browser keydown Control
+"##,
+        "keyup" => r##"
+agent-browser keyup - Release a key
+
+Usage: agent-browser keyup <key>
+
+Releases a key that was pressed with keydown.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser keyup Shift
+  agent-browser keyup Control
+"##,
+
+        // === Scroll ===
+        "scroll" => r##"
+agent-browser scroll - Scroll the page
+
+Usage: agent-browser scroll [direction] [amount]
+
+Scrolls the page in the specified direction.
+
+Arguments:
+  direction            up, down, left, right (default: down)
+  amount               Pixels to scroll (default: 300)
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser scroll
+  agent-browser scroll down 500
+  agent-browser scroll up 200
+  agent-browser scroll left 100
+"##,
+        "scrollintoview" | "scrollinto" => r##"
+agent-browser scrollintoview - Scroll element into view
+
+Usage: agent-browser scrollintoview <selector>
+
+Scrolls the page until the specified element is visible in the viewport.
+
+Aliases: scrollinto
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser scrollintoview "#footer"
+  agent-browser scrollintoview @e15
+"##,
+
+        // === Wait ===
+        "wait" => r##"
+agent-browser wait - Wait for condition
+
+Usage: agent-browser wait <selector|ms|option>
+
+Waits for an element to appear, a timeout, or other conditions.
+
+Modes:
+  <selector>           Wait for element to appear
+  <ms>                 Wait for specified milliseconds
+  --url <pattern>      Wait for URL to match pattern
+  --load <state>       Wait for load state (load, domcontentloaded, networkidle)
+  --fn <expression>    Wait for JavaScript expression to be truthy
+  --text <text>        Wait for text to appear on page
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser wait "#loading-spinner"
+  agent-browser wait 2000
+  agent-browser wait --url "**/dashboard"
+  agent-browser wait --load networkidle
+  agent-browser wait --fn "window.appReady === true"
+  agent-browser wait --text "Welcome back"
+"##,
+
+        // === Screenshot/PDF ===
+        "screenshot" => r##"
+agent-browser screenshot - Take a screenshot
+
+Usage: agent-browser screenshot [path]
+
+Captures a screenshot of the current page. If no path is provided,
+outputs base64-encoded image data.
+
+Options:
+  --full, -f           Capture full page (not just viewport)
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser screenshot
+  agent-browser screenshot ./screenshot.png
+  agent-browser screenshot --full ./full-page.png
+"##,
+        "pdf" => r##"
+agent-browser pdf - Save page as PDF
+
+Usage: agent-browser pdf <path>
+
+Saves the current page as a PDF file.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser pdf ./page.pdf
+  agent-browser pdf ~/Documents/report.pdf
+"##,
+
+        // === Snapshot ===
+        "snapshot" => r##"
+agent-browser snapshot - Get accessibility tree snapshot
+
+Usage: agent-browser snapshot [options]
+
+Returns an accessibility tree representation of the page with element
+references (like @e1, @e2) that can be used in subsequent commands.
+Designed for AI agents to understand page structure.
+
+Options:
+  -i, --interactive    Only include interactive elements
+  -c, --compact        Remove empty structural elements
+  -d, --depth <n>      Limit tree depth
+  -s, --selector <sel> Scope snapshot to CSS selector
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser snapshot
+  agent-browser snapshot -i
+  agent-browser snapshot --compact --depth 5
+  agent-browser snapshot -s "#main-content"
+"##,
+
+        // === Eval ===
+        "eval" => r##"
+agent-browser eval - Execute JavaScript
+
+Usage: agent-browser eval <script>
+
+Executes JavaScript code in the browser context and returns the result.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser eval "document.title"
+  agent-browser eval "window.location.href"
+  agent-browser eval "document.querySelectorAll('a').length"
+"##,
+
+        // === Close ===
+        "close" | "quit" | "exit" => r##"
+agent-browser close - Close the browser
+
+Usage: agent-browser close
+
+Closes the browser instance for the current session.
+
+Aliases: quit, exit
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser close
+  agent-browser close --session mysession
+"##,
+
+        // === Get ===
+        "get" => r##"
+agent-browser get - Retrieve information from elements or page
+
+Usage: agent-browser get <subcommand> [args]
+
+Retrieves various types of information from elements or the page.
+
+Subcommands:
+  text <selector>            Get text content of element
+  html <selector>            Get inner HTML of element
+  value <selector>           Get value of input element
+  attr <selector> <name>     Get attribute value
+  title                      Get page title
+  url                        Get current URL
+  count <selector>           Count matching elements
+  box <selector>             Get bounding box (x, y, width, height)
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser get text @e1
+  agent-browser get html "#content"
+  agent-browser get value "#email-input"
+  agent-browser get attr "#link" href
+  agent-browser get title
+  agent-browser get url
+  agent-browser get count "li.item"
+  agent-browser get box "#header"
+"##,
+
+        // === Is ===
+        "is" => r##"
+agent-browser is - Check element state
+
+Usage: agent-browser is <subcommand> <selector>
+
+Checks the state of an element and returns true/false.
+
+Subcommands:
+  visible <selector>   Check if element is visible
+  enabled <selector>   Check if element is enabled (not disabled)
+  checked <selector>   Check if checkbox/radio is checked
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser is visible "#modal"
+  agent-browser is enabled "#submit-btn"
+  agent-browser is checked "#agree-checkbox"
+"##,
+
+        // === Find ===
+        "find" => r##"
+agent-browser find - Find and interact with elements by locator
+
+Usage: agent-browser find <locator> <value> [action] [text]
+
+Finds elements using semantic locators and optionally performs an action.
+
+Locators:
+  role <role>              Find by ARIA role (--name <n>, --exact)
+  text <text>              Find by text content (--exact)
+  label <label>            Find by associated label (--exact)
+  placeholder <text>       Find by placeholder text (--exact)
+  alt <text>               Find by alt text (--exact)
+  title <text>             Find by title attribute (--exact)
+  testid <id>              Find by data-testid attribute
+  first <selector>         First matching element
+  last <selector>          Last matching element
+  nth <index> <selector>   Nth matching element (0-based)
+
+Actions (default: click):
+  click, fill, type, hover, focus, check, uncheck
+
+Options:
+  --name <name>        Filter role by accessible name
+  --exact              Require exact text match
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser find role button click --name Submit
+  agent-browser find text "Sign In" click
+  agent-browser find label "Email" fill "user@example.com"
+  agent-browser find placeholder "Search..." type "query"
+  agent-browser find testid "login-form" click
+  agent-browser find first "li.item" click
+  agent-browser find nth 2 ".card" hover
+"##,
+
+        // === Mouse ===
+        "mouse" => r##"
+agent-browser mouse - Low-level mouse operations
+
+Usage: agent-browser mouse <subcommand> [args]
+
+Performs low-level mouse operations for precise control.
+
+Subcommands:
+  move <x> <y>         Move mouse to coordinates
+  down [button]        Press mouse button (left, right, middle)
+  up [button]          Release mouse button
+  wheel <dy> [dx]      Scroll mouse wheel
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser mouse move 100 200
+  agent-browser mouse down
+  agent-browser mouse up
+  agent-browser mouse down right
+  agent-browser mouse wheel 100
+  agent-browser mouse wheel -50 0
+"##,
+
+        // === Set ===
+        "set" => r##"
+agent-browser set - Configure browser settings
+
+Usage: agent-browser set <setting> [args]
+
+Configures various browser settings and emulation options.
+
+Settings:
+  viewport <w> <h>           Set viewport size
+  device <name>              Emulate device (e.g., "iPhone 12")
+  geo <lat> <lng>            Set geolocation
+  offline [on|off]           Toggle offline mode
+  headers <json>             Set extra HTTP headers
+  credentials <user> <pass>  Set HTTP authentication
+  media [dark|light]         Set color scheme preference
+        [reduced-motion]     Enable reduced motion
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser set viewport 1920 1080
+  agent-browser set device "iPhone 12"
+  agent-browser set geo 37.7749 -122.4194
+  agent-browser set offline on
+  agent-browser set headers '{"X-Custom": "value"}'
+  agent-browser set credentials admin secret123
+  agent-browser set media dark
+  agent-browser set media light reduced-motion
+"##,
+
+        // === Network ===
+        "network" => r##"
+agent-browser network - Network interception and monitoring
+
+Usage: agent-browser network <subcommand> [args]
+
+Intercept, mock, or monitor network requests.
+
+Subcommands:
+  route <url> [options]      Intercept requests matching URL pattern
+    --abort                  Abort matching requests
+    --body <json>            Respond with custom body
+  unroute [url]              Remove route (all if no URL)
+  requests [options]         List captured requests
+    --clear                  Clear request log
+    --filter <pattern>       Filter by URL pattern
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser network route "**/api/*" --abort
+  agent-browser network route "**/data.json" --body '{"mock": true}'
+  agent-browser network unroute
+  agent-browser network requests
+  agent-browser network requests --filter "api"
+  agent-browser network requests --clear
+"##,
+
+        // === Storage ===
+        "storage" => r##"
+agent-browser storage - Manage web storage
+
+Usage: agent-browser storage <type> [operation] [key] [value]
+
+Manage localStorage and sessionStorage.
+
+Types:
+  local                localStorage
+  session              sessionStorage
+
+Operations:
+  get [key]            Get all storage or specific key
+  set <key> <value>    Set a key-value pair
+  clear                Clear all storage
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser storage local
+  agent-browser storage local get authToken
+  agent-browser storage local set theme "dark"
+  agent-browser storage local clear
+  agent-browser storage session get userId
+"##,
+
+        // === Cookies ===
+        "cookies" => r##"
+agent-browser cookies - Manage browser cookies
+
+Usage: agent-browser cookies [operation] [args]
+
+Manage browser cookies for the current context.
+
+Operations:
+  get                  Get all cookies (default)
+  set <name> <value>   Set a cookie
+  clear                Clear all cookies
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser cookies
+  agent-browser cookies get
+  agent-browser cookies set session_id "abc123"
+  agent-browser cookies clear
+"##,
+
+        // === Tabs ===
+        "tab" => r##"
+agent-browser tab - Manage browser tabs
+
+Usage: agent-browser tab [operation] [args]
+
+Manage browser tabs in the current window.
+
+Operations:
+  list                 List all tabs (default)
+  new [url]            Open new tab
+  close [index]        Close tab (current if no index)
+  <index>              Switch to tab by index
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser tab
+  agent-browser tab list
+  agent-browser tab new
+  agent-browser tab new https://example.com
+  agent-browser tab 2
+  agent-browser tab close
+  agent-browser tab close 1
+"##,
+
+        // === Window ===
+        "window" => r##"
+agent-browser window - Manage browser windows
+
+Usage: agent-browser window <operation>
+
+Manage browser windows.
+
+Operations:
+  new                  Open new browser window
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser window new
+"##,
+
+        // === Frame ===
+        "frame" => r##"
+agent-browser frame - Switch frame context
+
+Usage: agent-browser frame <selector|main>
+
+Switch to an iframe or back to the main frame.
+
+Arguments:
+  <selector>           CSS selector for iframe
+  main                 Switch back to main frame
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser frame "#embed-iframe"
+  agent-browser frame "iframe[name='content']"
+  agent-browser frame main
+"##,
+
+        // === Dialog ===
+        "dialog" => r##"
+agent-browser dialog - Handle browser dialogs
+
+Usage: agent-browser dialog <response> [text]
+
+Respond to browser dialogs (alert, confirm, prompt).
+
+Operations:
+  accept [text]        Accept dialog, optionally with prompt text
+  dismiss              Dismiss/cancel dialog
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser dialog accept
+  agent-browser dialog accept "my input"
+  agent-browser dialog dismiss
+"##,
+
+        // === Trace ===
+        "trace" => r##"
+agent-browser trace - Record execution trace
+
+Usage: agent-browser trace <operation> [path]
+
+Record a trace for debugging with Playwright Trace Viewer.
+
+Operations:
+  start [path]         Start recording trace
+  stop [path]          Stop recording and save trace
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser trace start
+  agent-browser trace start ./my-trace
+  agent-browser trace stop
+  agent-browser trace stop ./debug-trace.zip
+"##,
+
+        // === Console/Errors ===
+        "console" => r##"
+agent-browser console - View console logs
+
+Usage: agent-browser console [--clear]
+
+View browser console output (log, warn, error, info).
+
+Options:
+  --clear              Clear console log buffer
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser console
+  agent-browser console --clear
+"##,
+        "errors" => r##"
+agent-browser errors - View page errors
+
+Usage: agent-browser errors [--clear]
+
+View JavaScript errors and uncaught exceptions.
+
+Options:
+  --clear              Clear error buffer
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser errors
+  agent-browser errors --clear
+"##,
+
+        // === Highlight ===
+        "highlight" => r##"
+agent-browser highlight - Highlight an element
+
+Usage: agent-browser highlight <selector>
+
+Visually highlights an element on the page for debugging.
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser highlight "#target-element"
+  agent-browser highlight @e5
+"##,
+
+        // === State ===
+        "state" => r##"
+agent-browser state - Save/load browser state
+
+Usage: agent-browser state <operation> <path>
+
+Save or restore browser state (cookies, localStorage, sessionStorage).
+
+Operations:
+  save <path>          Save current state to file
+  load <path>          Load state from file
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser state save ./auth-state.json
+  agent-browser state load ./auth-state.json
+"##,
+
+        // === Session ===
+        "session" => r##"
+agent-browser session - Manage sessions
+
+Usage: agent-browser session [operation]
+
+Manage isolated browser sessions. Each session has its own browser
+instance with separate cookies, storage, and state.
+
+Operations:
+  (none)               Show current session name
+  list                 List all active sessions
+
+Environment:
+  AGENT_BROWSER_SESSION    Default session name
+
+Global Options:
+  --json               Output as JSON
+  --session <name>     Use specific session
+
+Examples:
+  agent-browser session
+  agent-browser session list
+  agent-browser --session test open example.com
+"##,
+
+        // === Install ===
+        "install" => r##"
+agent-browser install - Install browser binaries
+
+Usage: agent-browser install [--with-deps]
+
+Downloads and installs browser binaries required for automation.
+
+Options:
+  -d, --with-deps      Also install system dependencies (Linux only)
+
+Examples:
+  agent-browser install
+  agent-browser install --with-deps
+"##,
+
+        _ => return false,
+    };
+    println!("{}", help.trim());
+    true
+}
+
 pub fn print_help() {
    println!(
        r#"
@@ -231,6 +1189,8 @@ Snapshot Options:

 Options:
  --session <name>           Isolated session (or AGENT_BROWSER_SESSION env)
+  --headers <json>           HTTP headers scoped to URL's origin (for auth)
+  --executable-path <path>   Custom browser executable (or AGENT_BROWSER_EXECUTABLE_PATH)
  --json                     JSON output
  --full, -f                 Full page screenshot
  --headed                   Show browser window (not headless)
@@ -1,6 +1,6 @@
 {
  "name": "agent-browser",
-  "version": "0.4.2",
+  "version": "0.4.3",
  "description": "Headless browser automation CLI for AI agents",
  "type": "module",
  "main": "dist/daemon.js",
@@ -411,6 +411,12 @@ async function handleNavigate(
  browser: BrowserManager
 ): Promise<Response<NavigateData>> {
  const page = browser.getPage();
+
+  // If headers are provided, set up scoped headers for this origin
+  if (command.headers && Object.keys(command.headers).length > 0) {
+    await browser.setScopedHeaders(command.url, command.headers);
+  }
+
  await page.goto(command.url, {
    waitUntil: command.waitUntil ?? 'load',
  });
@@ -22,6 +22,16 @@ describe('BrowserManager', () => {
      const page = browser.getPage();
      expect(page).toBeDefined();
    });
+
+    it('should reject invalid executablePath', async () => {
+      const testBrowser = new BrowserManager();
+      await expect(
+        testBrowser.launch({
+          headless: true,
+          executablePath: '/nonexistent/path/to/chromium',
+        })
+      ).rejects.toThrow();
+    });
  });

  describe('navigation', () => {
@@ -294,4 +304,59 @@ describe('BrowserManager', () => {
      expect(h1).toBe('Example Domain');
    });
  });
+
+  describe('scoped headers', () => {
+    it('should register route for scoped headers', async () => {
+      // Test that setScopedHeaders doesn't throw and completes successfully
+      await browser.clearScopedHeaders();
+      await expect(
+        browser.setScopedHeaders('https://example.com', { 'X-Test': 'value' })
+      ).resolves.not.toThrow();
+      await browser.clearScopedHeaders();
+    });
+
+    it('should handle full URL origin', async () => {
+      await browser.clearScopedHeaders();
+      await expect(
+        browser.setScopedHeaders('https://api.example.com/path', { Authorization: 'Bearer token' })
+      ).resolves.not.toThrow();
+      await browser.clearScopedHeaders();
+    });
+
+    it('should handle hostname-only origin', async () => {
+      await browser.clearScopedHeaders();
+      await expect(
+        browser.setScopedHeaders('example.com', { 'X-Custom': 'value' })
+      ).resolves.not.toThrow();
+      await browser.clearScopedHeaders();
+    });
+
+    it('should clear scoped headers for specific origin', async () => {
+      await browser.clearScopedHeaders();
+      await browser.setScopedHeaders('https://example.com', { 'X-Test': 'value' });
+      await expect(browser.clearScopedHeaders('https://example.com')).resolves.not.toThrow();
+    });
+
+    it('should clear all scoped headers', async () => {
+      await browser.setScopedHeaders('https://example.com', { 'X-Test-1': 'value1' });
+      await browser.setScopedHeaders('https://example.org', { 'X-Test-2': 'value2' });
+      await expect(browser.clearScopedHeaders()).resolves.not.toThrow();
+    });
+
+    it('should replace headers when called twice for same origin', async () => {
+      await browser.clearScopedHeaders();
+      await browser.setScopedHeaders('https://example.com', { 'X-First': 'first' });
+      // Second call should replace, not add
+      await expect(
+        browser.setScopedHeaders('https://example.com', { 'X-Second': 'second' })
+      ).resolves.not.toThrow();
+      await browser.clearScopedHeaders();
+    });
+
+    it('should handle clearing non-existent origin gracefully', async () => {
+      await browser.clearScopedHeaders();
+      // Should not throw when clearing headers that were never set
+      await expect(browser.clearScopedHeaders('https://never-set.com')).resolves.not.toThrow();
+    });
+  });
 });
@@ -51,6 +51,7 @@ export class BrowserManager {
  private isRecordingHar: boolean = false;
  private refMap: RefMap = {};
  private lastSnapshot: string = '';
+  private scopedHeaderRoutes: Map<string, (route: Route) => Promise<void>> = new Map();

  /**
   * Check if browser is launched
@@ -439,7 +440,7 @@ export class BrowserManager {
  }

  /**
-   * Set extra HTTP headers
+   * Set extra HTTP headers (global - all requests)
   */
  async setExtraHeaders(headers: Record<string, string>): Promise<void> {
    const context = this.contexts[0];
@@ -448,6 +449,76 @@ export class BrowserManager {
    }
  }

+  /**
+   * Set scoped HTTP headers (only for requests matching the origin)
+   * Uses route interception to add headers only to matching requests
+   */
+  async setScopedHeaders(origin: string, headers: Record<string, string>): Promise<void> {
+    const page = this.getPage();
+
+    // Build URL pattern from origin (e.g., "api.example.com" -> "**://api.example.com/**")
+    // Handle both full URLs and just hostnames
+    let urlPattern: string;
+    try {
+      const url = new URL(origin.startsWith('http') ? origin : `https://${origin}`);
+      // Match any protocol, the host, and any path
+      urlPattern = `**://${url.host}/**`;
+    } catch {
+      // If parsing fails, treat as hostname pattern
+      urlPattern = `**://${origin}/**`;
+    }
+
+    // Remove existing route for this origin if any
+    const existingHandler = this.scopedHeaderRoutes.get(urlPattern);
+    if (existingHandler) {
+      await page.unroute(urlPattern, existingHandler);
+    }
+
+    // Create handler that adds headers to matching requests
+    const handler = async (route: Route) => {
+      const requestHeaders = route.request().headers();
+      await route.continue({
+        headers: {
+          ...requestHeaders,
+          ...headers,
+        },
+      });
+    };
+
+    // Store and register the route
+    this.scopedHeaderRoutes.set(urlPattern, handler);
+    await page.route(urlPattern, handler);
+  }
+
+  /**
+   * Clear scoped headers for an origin (or all if no origin specified)
+   */
+  async clearScopedHeaders(origin?: string): Promise<void> {
+    const page = this.getPage();
+
+    if (origin) {
+      let urlPattern: string;
+      try {
+        const url = new URL(origin.startsWith('http') ? origin : `https://${origin}`);
+        urlPattern = `**://${url.host}/**`;
+      } catch {
+        urlPattern = `**://${origin}/**`;
+      }
+
+      const handler = this.scopedHeaderRoutes.get(urlPattern);
+      if (handler) {
+        await page.unroute(urlPattern, handler);
+        this.scopedHeaderRoutes.delete(urlPattern);
+      }
+    } else {
+      // Clear all scoped header routes
+      for (const [pattern, handler] of this.scopedHeaderRoutes) {
+        await page.unroute(pattern, handler);
+      }
+      this.scopedHeaderRoutes.clear();
+    }
+  }
+
  /**
   * Start tracing
   */
@@ -520,11 +591,13 @@ export class BrowserManager {
    // Launch browser
    this.browser = await launcher.launch({
      headless: options.headless ?? true,
+      executablePath: options.executablePath,
    });

-    // Create context with viewport
+    // Create context with viewport and optional headers
    const context = await this.browser.newContext({
      viewport: options.viewport ?? { width: 1280, height: 720 },
+      extraHTTPHeaders: options.headers,
    });

    // Set default timeout to 10 seconds (Playwright default is 30s)
@@ -158,7 +158,12 @@ export async function startDaemon(): Promise<void> {
            parseResult.command.action !== 'launch' &&
            parseResult.command.action !== 'close'
          ) {
-            await browser.launch({ id: 'auto', action: 'launch', headless: true });
+            await browser.launch({
+              id: 'auto',
+              action: 'launch',
+              headless: true,
+              executablePath: process.env.AGENT_BROWSER_EXECUTABLE_PATH,
+            });
          }

          // Handle close command specially
@@ -12,12 +12,15 @@ export interface LaunchCommand extends BaseCommand {
  headless?: boolean;
  viewport?: { width: number; height: number };
  browser?: 'chromium' | 'firefox' | 'webkit';
+  headers?: Record<string, string>;
+  executablePath?: string;
 }

 export interface NavigateCommand extends BaseCommand {
  action: 'navigate';
  url: string;
  waitUntil?: 'load' | 'domcontentloaded' | 'networkidle';
+  headers?: Record<string, string>;
 }

 export interface ClickCommand extends BaseCommand {
@@ -0,0 +1,85 @@
+/**
+ * Integration test for @sparticuz/chromium compatibility
+ * This tests the executablePath option with a serverless-optimized Chromium build
+ *
+ * Note: @sparticuz/chromium only works on Linux (designed for AWS Lambda).
+ * This test will skip on non-Linux platforms.
+ */
+import { describe, it, expect, afterAll } from 'vitest';
+import { BrowserManager } from '../src/browser.js';
+import * as os from 'os';
+
+const isLinux = os.platform() === 'linux';
+
+// Only run if @sparticuz/chromium is available AND we're on Linux
+const canRunTest = await (async () => {
+  if (!isLinux) {
+    console.log('Skipping @sparticuz/chromium test: only runs on Linux');
+    return false;
+  }
+  try {
+    await import('@sparticuz/chromium');
+    return true;
+  } catch {
+    console.log('Skipping @sparticuz/chromium test: package not installed');
+    return false;
+  }
+})();
+
+describe.skipIf(!canRunTest)('Serverless Chromium Integration', () => {
+  let browser: BrowserManager;
+  let chromiumPath: string;
+
+  it('should get executable path from @sparticuz/chromium', async () => {
+    const chromium = await import('@sparticuz/chromium');
+    chromiumPath = await chromium.default.executablePath();
+    expect(chromiumPath).toBeTruthy();
+    expect(typeof chromiumPath).toBe('string');
+    console.log('Chromium executable path:', chromiumPath);
+  });
+
+  it('should launch browser with custom executablePath', async () => {
+    const chromium = await import('@sparticuz/chromium');
+    chromiumPath = await chromium.default.executablePath();
+
+    browser = new BrowserManager();
+    await browser.launch({
+      headless: true,
+      executablePath: chromiumPath,
+    });
+
+    expect(browser.isLaunched()).toBe(true);
+  });
+
+  it('should navigate to a page', async () => {
+    const page = browser.getPage();
+    await page.goto('https://example.com');
+    expect(page.url()).toBe('https://example.com/');
+  });
+
+  it('should get page title', async () => {
+    const page = browser.getPage();
+    const title = await page.title();
+    expect(title).toBe('Example Domain');
+  });
+
+  it('should take snapshot with refs', async () => {
+    const { tree, refs } = await browser.getSnapshot();
+    expect(tree).toContain('Example Domain');
+    expect(typeof refs).toBe('object');
+    expect(Object.keys(refs).length).toBeGreaterThan(0);
+  });
+
+  it('should take screenshot', async () => {
+    const page = browser.getPage();
+    const buffer = await page.screenshot();
+    expect(buffer).toBeInstanceOf(Buffer);
+    expect(buffer.length).toBeGreaterThan(0);
+  });
+
+  afterAll(async () => {
+    if (browser?.isLaunched()) {
+      await browser.close();
+    }
+  });
+});
@@ -3,7 +3,7 @@ import { defineConfig } from 'vitest/config';
 export default defineConfig({
  test: {
    globals: true,
-    include: ['src/**/*.test.ts'],
+    include: ['src/**/*.test.ts', 'test/**/*.test.ts'],
    testTimeout: 30000,
  },
 });
Author	SHA1	Message	Date
Chris Tate	068fc74a9c	Merge main into ctate/custom-headers	2026-01-12 11:57:43 -06:00
Chris Tate	4f6fd8ec5c	support serverless environments (#29 ) * add --executable-path * tests * test vercel * fixes	2026-01-12 11:41:25 -06:00
Chris Tate	4e5ff20078	better parsing	2026-01-12 11:40:52 -06:00
Chris Tate	03e266ee38	add tests	2026-01-12 11:26:45 -06:00
Chris Tate	8c412197ad	add custom headers via --headers	2026-01-12 11:22:46 -06:00
Chris Tate	3cd0ab468f	add sub --help flag (#27 )	2026-01-12 10:52:31 -06:00
Chris Tate	a4fcc1c198	fix windows bug (#26 ) * fix windows bug * test windows * address feedback	2026-01-12 10:33:21 -06:00
Chris Tate	574037080c	0.4.3 (#20 )	2026-01-12 01:24:48 -06:00
Chris Tate	278466764b	fix readme + add missing wait flags (#19 ) * fix inaccuracies * fix wait * address feedback	2026-01-12 01:22:10 -06:00