GSOC 2026 Inquiry: Project #10 Agentic Dora

[23garyd]

I have a few questions to align on before sketching out a proposal:

• Do you have a preference on the agent framework , or should I choose based on which handles custom tool-calling with the least overhead?

• Simulation Scope: For the PyBullet environment, would you prefer a manipulator arm (for pick/place tasks) or a mobile base (for navigation) for the initial demo?

• Mocking: For the offline testing, would you prefer a deterministic rule-based mock agent, or a small local LLM to better simulate real-world latency/errors?

[bobdingAI]

Great questions — here are my thoughts:
Agent: We don't have a strong preference on a specific framework. An open-source option with a small footprint that handles custom tool-calling cleanly would be ideal. Feel free to pick one, just include a brief justification in your proposal for why you chose it.
Simulation: Either a manipulator arm or mobile base would work for the initial demo. Pick whichever you're more comfortable with or think makes a stronger showcase, and outline your reasoning in the proposal.
Mocking: I'd encourage you to research both approaches and come back with a recommendation. Propose what you think is the better fit and explain why.

[23garyd]

what are differences between https://github.com/dora-rs/adora and https://github.com/dora-rs/dora ?

[bobdingAI]

https://github.com/dora-rs/adora is experimental project. I have change the links to use https://github.com/dora-rs/dora.