Troubleshooting: Browser Automation

Chrome / Chromium Not Found

Triggerfish uses puppeteer-core (not bundled Chromium) and auto-detects Chrome or Chromium on your system. If no browser is found, browser tools will fail with a launch error.

Detection paths by platform

Linux:

• /usr/bin/chromium
• /usr/bin/chromium-browser
• /usr/bin/google-chrome
• /usr/bin/google-chrome-stable
• /snap/bin/chromium
• /usr/bin/brave
• /usr/bin/brave-browser
• Flatpak: com.google.Chrome, org.chromium.Chromium, com.brave.Browser

macOS:

• /Applications/Google Chrome.app/Contents/MacOS/Google Chrome
• /Applications/Brave Browser.app/Contents/MacOS/Brave Browser
• /Applications/Chromium.app/Contents/MacOS/Chromium

Windows:

• %PROGRAMFILES%\Google\Chrome\Application\chrome.exe
• %PROGRAMFILES(X86)%\Google\Chrome\Application\chrome.exe
• %LOCALAPPDATA%\Google\Chrome\Application\chrome.exe

Installing a browser

# Debian/Ubuntu
sudo apt install chromium-browser

# Fedora
sudo dnf install chromium

# macOS
brew install --cask google-chrome

# Or install Brave, which is also detected

Manual path override

If your browser is installed in a non-standard location, you can set the path. Contact the project for the exact config key (this is currently set via the browser manager configuration).

---

Launch Failures

"Direct Chrome process launch failed"

Triggerfish launches Chrome in headless mode via Deno.Command. If the process fails to start:

1. Binary is not executable. Check file permissions.
2. Missing shared libraries. On minimal Linux installs (containers, WSL), Chrome may need additional libraries:

   # Debian/Ubuntu
   sudo apt install libnss3 libatk1.0-0 libatk-bridge2.0-0 libcups2 libxcomposite1 libxrandr2 libgbm1 libpango-1.0-0 libcairo2 libasound2

3. No display server. Chrome headless does not need X11/Wayland, but some Chrome versions still try to load display-related libraries.

Flatpak Chrome

If Chrome is installed as a Flatpak package, Triggerfish creates a wrapper script that calls flatpak run with the appropriate arguments.

Flatpak wrapper script file write failed
Flatpak Chrome process launch failed
Flatpak Chrome launch failed

If the wrapper script fails:

• Check that /usr/bin/flatpak or /usr/local/bin/flatpak exists
• Check that the Flatpak app ID is correct (run flatpak list to see installed apps)
• The wrapper script is written to a temp directory. If the temp directory is not writable, the write fails.

CDP endpoint not ready

After launching Chrome, Triggerfish polls the Chrome DevTools Protocol (CDP) endpoint to establish a connection. Default timeout is 30 seconds with 200ms polling interval.

CDP endpoint on port <port> not ready after <timeout>ms

This means Chrome started but did not open the CDP port in time. Causes:

• Chrome is loading slowly (resource-constrained system)
• Another Chrome instance is using the same debugging port
• Chrome crashed during startup (check Chrome's own output)

---

Navigation Issues

"Navigation blocked by domain policy"

The browser tools apply the same SSRF protection as web_fetch. URLs pointing to private IP addresses are blocked:

Navigation blocked by domain policy: http://192.168.1.1/admin

This is intentional security enforcement. The browser cannot access:

• localhost / 127.0.0.1
• Private networks (10.x.x.x, 172.16-31.x.x, 192.168.x.x)
• Link-local addresses (169.254.x.x)

There is no way to disable this check.

"Invalid URL"

The URL is malformed. Browser navigation requires a full URL with protocol:

# Wrong
browser_navigate google.com

# Right
browser_navigate https://google.com

Navigation timeout

Navigation failed: Timeout

The page took too long to load. This is typically a slow server or a page that never finishes loading (infinite redirects, stuck JavaScript).

---

Page Interaction Issues

"Click failed", "Type failed", "Select failed"

These errors include the CSS selector that failed:

Click failed on ".submit-button": Node not found
Type failed on "#email": Node not found

The selector did not match any element on the page. Common causes:

• The page has not finished loading yet
• The element is inside an iframe (selectors do not cross iframe boundaries)
• The selector is wrong (dynamic class names, shadow DOM)

"Snapshot failed"

The page snapshot (DOM extraction for context) failed. This can happen if:

• The page has no content (blank page)
• JavaScript errors prevent DOM access
• The page navigated away during snapshot capture

"Scroll failed"

Usually happens on pages with custom scroll containers. The scroll command targets the main document viewport.

---

Profile Isolation

Browser profiles are isolated per agent. Each agent gets its own Chrome profile directory under the profile base directory. This means:

• Login sessions are not shared between agents
• Cookies, local storage, and cache are per-agent
• Classification-aware access controls prevent cross-contamination

If you see unexpected profile behavior, the profile directory may be corrupted. Delete it and let Triggerfish create a fresh one on the next browser launch.