Google Summer of Code 2026 Project List

[bdmorgan]

As I mentioned earlier this week, we have updated our project list based on current needs and individual maintainers' own interests. Project ideas have been documented as issues in the repo. The current projects are as follows:

• Terminal-Integrated Performance & Memory Investigation Companion - Mentor: @gsquared94
• Behavioral evals, Quality, and the OSS Community - Mentor: @gundermanc
• Optimize Test Suite Output Logging - Mentor: @scidomino
• Long-Context & Complex Reasoning Coding Evaluation Dataset - Mentor: @akh64bit

Please reply to those issues directly if you're interested, want to ask questions, or want to make proposals. Thank you for your interest in Gemini CLI and Google Summer of Code 2026!

[Aaxhirrr]

We really appreciate it, thank you !

[hscecoder]

Is there any pre-defined template for proposals?

[Aaxhirrr]

Well GSoC has a default template as long as i know, and organizations give out their own templates sometimes. As far as i know, gemini cli has not provided a template, so the best case scenario is to stick with the GSoC's default.

[likhith-adithya]

Thank you.

[ishanmitra]

@bdmorgan After submitting a proposal on the GSOC platform, is there a possibility of receiving a review from the assigned mentor? Thank you. ✨

[anxkhn]

Projects lists are released finally! ✨

[PoleacLiviuStefan]

Thank you for the list

[ghost]

Thank you for this list! :)

[PoleacLiviuStefan]

So based on this issues we should submit proposal with our implementation on GSoC platform?

[bdmorgan]

Yes I think that is how it should work. This is the first summer we have participated so we are still learning ourselves.

[7hokerz]

😄😄😄

[Muawiya-contact]

Let's dive in it!

[Anjaligarhwal]

Hey @bdmorgan , thanks for sharing this project list, it’s genuinely exciting to go through
The focus on using LLMs for performance and debugging workflows stands out, especially shifting away from step-heavy debugging toward higher level analysis. The behavioral evals and test logging work also feel like high-impact areas for both developer experience and reliability. The long-context dataset direction is particularly interesting, moving toward realistic, multi-repo problems seems like the right step for evaluating true agent capability.

I’ll dig into the individual issues and try to contribute where I can. Looking forward to learning and being involved here.

[arlamanasa2007-madhu]

Thank you

[vidhip222]

Hi @bdmorgan, thanks for sharing this. These projects look very impactful!

I’ve already started exploring the repo and I’m planning to stay consistent by contributing daily to build context around the CLI and its architecture.

I had a quick question for proposals, is early feedback via issue discussions the best path, or should we come in with more concrete drafts/PRs first?

[Jyz922]

Thank you!

[Aarchi-07]

Hi @bdmorgan!
I'm interested in this project. I've been contributing to the test infrastructure already, e.g. #20439 (merged by @scidomino) fixed hook propagation in the retry path, and #22037 added missing temp directory cleanup in test afterEach hook. Would love to work on this!

[manishalakshminarayanan31]

Hello, I am interested in working on this project for GSoC 2026.

I have experience in Python, GitHub, and basic software development.
I would like to learn more about this project and start contributing.

Could you please guide me on beginner issues or tasks to start with?

Thank you.

[JayadityaGit]

There is no such thing as "beginner level issues" either you go through the issue tracker or you use the project and find an issue yourself and start working on it.

Remember that, it is not about "How much you know" it is about "What can you see or spot and have insight in that moment"

Best of Luck !

[S6S-sastrad]

As I mentioned earlier this week, we have updated our project list based on current needs and individual maintainers' own interests. Project ideas have been documented as issues in the repo. The current projects are as follows:

• Terminal-Integrated Performance & Memory Investigation Companion - Mentor: @gsquared94
• Behavioral evals, Quality, and the OSS Community - Mentor: @gundermanc
• Optimize Test Suite Output Logging - Mentor: @scidomino
• Long-Context & Complex Reasoning Coding Evaluation Dataset - Mentor: @akh64bit

Please reply to those issues directly if you're interested, want to ask questions, or want to make proposals. Thank you for your interest in Gemini CLI and Google Summer of Code 2026!

[campox747]

Great thanks for the update! Looking to contribute to these issues

[Abatye-andarg]

As I mentioned earlier this week, we have updated our project list based on current needs and individual maintainers' own interests. Project ideas have been documented as issues in the repo. The current projects are as follows:

• Terminal-Integrated Performance & Memory Investigation Companion - Mentor: @gsquared94
• Behavioral evals, Quality, and the OSS Community - Mentor: @gundermanc
• Optimize Test Suite Output Logging - Mentor: @scidomino
• Long-Context & Complex Reasoning Coding Evaluation Dataset - Mentor: @akh64bit

Please reply to those issues directly if you're interested, want to ask questions, or want to make proposals. Thank you for your interest in Gemini CLI and Google Summer of Code 2026!

Interested

[hscecoder]

Is there any pre-defined template for proposals?

[Rishabh-Sahni-0809]

Hello @bdmorgan , A very pleasure to meet you...
I have an Question So Should we apply through the proposals based on the above issues or there is some prerequisite tasks or we can start contributing there

[priti03-star]

Hi @akh64bit, I’m a Python developer interested in the 'Long-Context & Complex Reasoning' project. Are there any specific Python libraries you recommend I familiarize myself with for this dataset work?

[vanshksingh]

Hi @bdmorgan - I'm Vansh Kumar Singh, GSoC 2025 with Google DeepMind. I'd like to propose an unlisted GSoC project and wanted to check scope fit before submitting formally.

The idea: native first-class Ollama/local LLM backend support for Gemini CLI.

The community need is documented. Issue #2168 (open since June 2025) reports that GOOGLE_GEMINI_BASE_URL is not propagated into sandbox containers, breaking any local backend setup in sandbox mode. Issue #15430 (labeled help-wanted, priority/p2) explicitly requests GOOGLE_GEMINI_BASE_URL be respected as a top-priority override. In February 2026, a developer published a community workaround using LiteLLM Proxy as middleware - requiring three separate running processes and a manually-maintained model alias YAML that breaks on every Gemini CLI version update (dev.to/polar3130/using-gemini-cli-with-a-local-llm-5f5l).

My proposal: a pluggable LLMBackend interface in packages/core/src/api/, an OllamaBackend implementation that translates between Gemini API and Ollama REST formats, a --backend flag and settings.json config block, and the sandbox environment fix for Issue #2168. I'm also scoping in Gemma model support - Gemma 3 supports tool calling and is available via both the Gemini API and Ollama, so this brings Google's own open models into the Gemini CLI ecosystem natively.

I have direct Ollama experience and months of Gemini API integration from my DeepMind work.

I will be submitting my proposal for this issue. I hope the old sees it fit to be accepted , maybe as the long lost 5th idea ;)

Also ! Amazing work from the team, looking forward to see how it pans out !

GitHub Gist | GSoC 2025

[Rudra2122]

Hi @bdmorgan , I’m very interested in this project, especially the focus on combining memory analysis, profiling, and LLM-based reasoning within a CLI workflow.

From my understanding, the key challenge is not just collecting data (heap snapshots, CPU profiles), but structuring it in a way that makes root cause analysis reliable and actionable.

I’m thinking of approaching this in three layers:

snapshot + profiling data collection
signal extraction (growth patterns, retained objects, hotspots)
LLM-assisted summarization for root cause suggestions

I have experience building low-latency systems and working with profiling and observability tools, so this aligns closely with my background.

I had a couple of questions:

1. Would you expect the initial focus to prioritize memory (heap snapshots + diffing) before CPU profiling?
2. Are there existing abstractions in Gemini CLI for integrating structured diagnostic signals into LLM prompts?

I’m planning to prototype a small snapshot diff + analysis tool to explore this further. Would love any guidance on where to start within the repo.

[krishnakumarsethi]

Hi @bdmorgan excited for this project. Terminal integrated performance and memory investigation companion @gsquared94 .

[umachandola249]

Hello, I’m Uma, a first-year BS Data Science and Programming student .
I’m very interested in the project “Behavioral evals, Quality, and the OSS Community”.

I have experience with Python, SQL, Excel, and Power BI, which I’ve used for data analysis and visualization. I’m confident I can apply these skills to design evaluation metrics, analyze contribution quality, and build useful insights for the OSS community.

I would love to work under the guidance of @gundermanc and learn more about how to contribute effectively.

I also want to thank the Gemini CLI maintainers, mentors, and collaborators for putting together such a wonderful opportunity for students and learners like me.

[CogitationOps]

Hi @bdmorgan
I took a look at the project and have already contributed to a few issues in the Gemini CLI. I’m interested in working on the above proposed projects.

What would you recommend as the best approach from here? Should I start preparing something specifically aligned with these projects, or continue contributing to related CLI issues if the goal is to build toward a GSOC proposal?

[sanju-aiml]

As I mentioned earlier this week, we have updated our project list based on current needs and individual maintainers' own interests. Project ideas have been documented as issues in the repo. The current projects are as follows:

• Terminal-Integrated Performance & Memory Investigation Companion - Mentor: @gsquared94
• Behavioral evals, Quality, and the OSS Community - Mentor: @gundermanc
• Optimize Test Suite Output Logging - Mentor: @scidomino
• Long-Context & Complex Reasoning Coding Evaluation Dataset - Mentor: @akh64bit

Please reply to those issues directly if you're interested, want to ask questions, or want to make proposals. Thank you for your interest in Gemini CLI and Google Summer of Code 2026!

.

[sanju-aiml]

Can anyone guide me https://www.linkedin.com/in/sanjay-kumar-357011369?utm_source=share_via&utm_content=profile&utm_medium=member_android this is my LinkedIn profile iam new to here gsoc

[phaniratan1234]

Hi @bdmorgan - Just want to check. do you consider any new proposal other than the 4 mentioned above and allocate mentor bandwitdth, if it's a real pain point ?

Thank you.

[ishanmitra]

@bdmorgan I have been confused about the length and duration of the project for three out of the four projects. Any incorrect information in the proposal could become subject to disqualification. I applied for Optimize Test Suite Logging and would like to ensure what the length of the project is going to be. Thank you ✨