ntroduction: interested in the Agentic Dora idea

[mac146]

Hi everyone,

I'm Mayank, an electronics and communication student from Delhi (graduating in 2028). I work part-time on an agentic AI product for finance, where I built a pipeline execution engine and a retrieval (RAG) system. I also build a NestJS backend for a startup. I mostly write Python and TypeScript, and I'm learning Rust.

I'm interested in the "Agentic Dora" idea and I'm thinking about applying to GSoC next cycle. To get started I cloned the finished gsoc2026-dora-agentic-robot project and ran its tests against dora 1.1.0. The 191 unit tests passed, and so did the 12 real dataflow tests (loopback and failure recovery). One test, test_fk_matches_mujoco_across_configs, fails on a fresh clone because the robot meshes are not downloaded yet, while the other mesh tests skip in that case. I'd be happy to send a small PR for that if it's welcome.

I read the "Future work" section of the final report and have two questions:

Will Agentic Dora be offered again next year? If yes, which item would you like a new contributor to start with: the perception loop, a second robot, or something else?
Is there a good first issue in dora itself for someone who knows Python better than Rust?
Some of my work: I built an MCP server called IssueWatcher (github.com/mac146/IssueWatcher), and I have an open PR in google-gemini/genai-processors that I'm working through review on.

Thanks for reading!
Mayank