-
Notifications
You must be signed in to change notification settings - Fork 37
Add option to await browser tasks via env var #19
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
Hey @samanthavbarron, thanks for the kudos and playing with this mcp server! Sounds great, especially your deployment in a n8n environment. We didn't settle for a license of this repo and a related contribution handling, so please bear with us to figure this out in the next days. We'll check your implementation and your hint regarding possibly more elegant options on this behavior. The flag is certainly a valid option, another one might be a parameter to the mcp call or another mcp tool altogether, let's see. If you have further remarks or concerns, feel free to comment over here or shoot us an email to info at cobrowser.xyz. |
Hey @samanthavbarron, thanks for the PR and the suggestion! The flag is a great idea—In addition I’d recommend adding it as a parameter to the existing mcp tool rather than creating a new tool. This approach offers greater flexibility on the client side, accommodating various integration scenarios depending on the specific MCP client used (e.g., cursor, MCP inspector), since each provides different interactions with MCP servers. |
Hey, sorry to comment in the middle. |
Hey @wtberry — I think you made a great call. When we're awaiting the browser_use task itself, the response shouldn't include the job ID. Since there are no further updates at that point, the job ID doesn't really serve a purpose. That said, I do think keeping the status field is still useful and makes sense. |
@michiosw Sounds good, I can create a PR to make that change once this PR is approved and merged. |
Hey @wtberry probably today or tomorrow. I will need to add some adjustments but then it should be ready to go so you can create your PR. Make sure to check our contribution docs. |
Hey @wtberry and @samanthavbarron , I added necessary changes and will merge the PR later today. Thank you for your contribution! |
Implemented PATIENT mode that allows browser tasks to complete synchronously, with proper error handling and configuration. When enabled, browser_use tool returns complete results directly. Updated tool descriptions to clearly indicate behavior in each mode.
78d0e23
to
1aeec10
Compare
@michiosw Thanks! I will go ahead and create PR when ready! |
@wtberry I think this PR covers all the features you suggested. However, feel free to review and open another PR if you see any room for improvement! |
🎉 This PR is included in version 1.0.0 🎉 The release is available on GitHub release Your semantic-release bot 📦🚀 |
First off, after trying a bunch of similar tools, this is the best I've found for my use case so far. Great job! 🎉
This PR adds an environment variable option to wait for API calls to complete before returning the task. There's probably a better way of doing this.
My motivation for doing this is that I'm using this MCP server in an n8n workflow, however in n8n there's not a way to insert a delay between tool calls when using the agent node (AFAIK).