Troubleshooting

Having issues with BrowseGenius? Start here for quick resolutions and diagnostics.

Common problems

Ensure you’ve saved the key in the popup settings pane.
Keys are scoped per browser profile; re-enter if you switched profiles.
Verify your OpenAI account has access to the selected model (GPT-4.1, GPT-4o, o1, etc.).

Capture fails with “Unable to capture the current tab”

Chrome blocks tab capture on internal pages (chrome://, chromewebstore://). Switch to a regular HTTPS page.
Check that the extension has activeTab permission (reloading the extension after installation helps).
Try refreshing the target tab to re-inject content scripts.

Suite stops instantly

The active tab may be a PDF or extension page that disallows debugger attachment.
Inspect DevTools console logs (chrome://extensions → BrowseGenius → Inspect views).
Confirm that latency to OpenAI is not timing out (retry with a simpler flow).

GPT keeps choosing invalid actions

Re-capture updated screens to refresh DOM IDs.
Add clearer expectations in capture notes so the model knows what success looks like.
Reduce noise by removing optional captures that don’t influence the flow.

Reports fail to download

The browser may block automatic downloads. Check the downloads icon for prompts.
On first download, Chrome asks for permission; click Allow.
Ensure corporate policies or endpoint managers are not preventing chrome.downloads.

Diagnostic tools

Task history: The popup lists raw prompts and responses for each action.
DevTools logs: chrome://extensions → BrowseGenius → service worker console surfaces orchestrator errors.
OpenPipe telemetry: Supply an OpenPipe key to cross-check LLM prompts and completions.

Resetting state

Use the options menu (gear icon) to clear stored API keys.
Delete individual captures or press Reset (coming soon) to wipe state slices.
As a last resort, remove and re-load the extension; persisted data exists only in browser storage.