hermes - 💡(How to fix) Fix browser_tool.py: expose agent-browser upload command for file input elements [2 comments, 2 participants]

TL;DR

The proposed solution involves adding a browser_upload_file function to the Hermes browser toolset to enable file uploads via the agent-browser upload command.

Guidance

• Review the implementation of browser_upload_file in tools/browser_tool.py to ensure it correctly delegates to agent-browser upload <ref> <path>.
• Verify that the schema entry in BROWSER_TOOL_SCHEMAS is correctly defined for the new browser_upload_file function.
• Test the new function with various CSS selectors and local file paths to ensure it works as expected.
• Consider adding error handling for cases where the file upload fails or the input[type=file] element is not found.

Example

def browser_upload_file(ref, path):
    # Delegate to agent-browser upload command
    return agent_browser.upload(ref, path)

Notes

The proposed solution relies on the agent-browser tool already supporting the upload command, so no additional changes are needed outside of the Hermes browser toolset.

Recommendation

Apply the proposed upstream change by adding the browser_upload_file function and tool registration to the main branch, as it provides a minimal and effective solution to the problem.