Google Summer of Code 2026 Project List #23391
Replies: 59 comments 19 replies
-
|
We really appreciate it, thank you ! |
Beta Was this translation helpful? Give feedback.
-
|
Is there any pre-defined template for proposals? |
Beta Was this translation helpful? Give feedback.
-
|
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. |
Beta Was this translation helpful? Give feedback.
-
|
Hi, could you please clarify the project sizes for GSoC (small/medium/large) and how we should choose the correct one while filling the portal? Having a clear idea about this is very important to avoid mistakes in the application. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you. |
Beta Was this translation helpful? Give feedback.
-
|
@bdmorgan After submitting a proposal on the GSOC platform, is there a possibility of receiving a review from the assigned mentor? Thank you. ✨ |
Beta Was this translation helpful? Give feedback.
-
|
Projects lists are released finally! ✨ |
Beta Was this translation helpful? Give feedback.
-
|
Thank you for the list |
Beta Was this translation helpful? Give feedback.
-
|
Thank you for this list! :) |
Beta Was this translation helpful? Give feedback.
-
|
So based on this issues we should submit proposal with our implementation on GSoC platform? |
Beta Was this translation helpful? Give feedback.
-
|
Yes I think that is how it should work. This is the first summer we have participated so we are still learning ourselves. |
Beta Was this translation helpful? Give feedback.
-
|
😄😄😄 |
Beta Was this translation helpful? Give feedback.
-
|
Let's dive in it! |
Beta Was this translation helpful? Give feedback.
-
|
Hey @bdmorgan , thanks for sharing this project list, it’s genuinely exciting to go through I’ll dig into the individual issues and try to contribute where I can. Looking forward to learning and being involved here. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you |
Beta Was this translation helpful? Give feedback.
-
|
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? |
Beta Was this translation helpful? Give feedback.
-
|
Thank you! |
Beta Was this translation helpful? Give feedback.
-
|
Hi @bdmorgan! |
Beta Was this translation helpful? Give feedback.
-
|
Hello, I am interested in working on this project for GSoC 2026. I have experience in Python, GitHub, and basic software development. Could you please guide me on beginner issues or tasks to start with? Thank you. |
Beta Was this translation helpful? Give feedback.
-
|
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 ! |
Beta Was this translation helpful? Give feedback.
-
|
Hi @bdmorgan, I hope you're doing well. I wanted to share a quick concern about communication with my mentor, @gsquared94. I understand that everyone has a busy schedule, and I truly appreciate the time and effort you put into supporting us. However, I’ve been finding it a bit difficult to get responses, which is making it harder for me to fully benefit from the mentorship experience (like the proposal review). Thank you for your support and understanding. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you |
Beta Was this translation helpful? Give feedback.
-
|
Hello @scidomino , |
Beta Was this translation helpful? Give feedback.
-
|
Hi, I'm Yahia. I've just submitted a GSoC proposal to build a Performance Dashboard for the CLI. I'm very interested in how the CLI handles latency currently. Is there any specific part of the codebase you recommend I look into first to understand the current telemetry? |
Beta Was this translation helpful? Give feedback.
-
|
Thank you for this list! :) |
Beta Was this translation helpful? Give feedback.
-
|
Hello, @bdmorgan The problem I want to solve: In long coding sessions (30+ turns), the majority of tokens sent to the API are stale, old read_file results, obsolete test logs, early shell output that the model never revisits. Today This isn't a guess, I've been profilling real coding agent sessions as part of my research, and the pattern is remarkably consistent: tool outputs follow a write-once-read-once pattern, with 56-76% of context going cold within 2-3 turns. System prompts stay hot regardless of session length The proposal: Role aware tiered context management. Classify segments by role tag + turn age (zero overhead, tags are already in the chat template), then keep recent turns full, summarize aging outputs, stub old history with metadata. Extends Happy to write up a full issue with technical details, integration points, and timeline. This work is part of my broader research on agentic LLM workloads optimization, so I'm committed to maintaining and extending it well beyond the GSoC period. If there's interest from a potential mentor, please feel free to reach out. |
Beta Was this translation helpful? Give feedback.
-
|
Hi @bdmorgan, Thank you for sharing the GSoC 2026 project list — the ideas look really impactful. I’m particularly interested in "Optimize Test Suite Output Logging", as it aligns well with my experience in backend development, Node.js, logging systems, and performance optimization. I’ve started exploring the repository and related issues, and I plan to contribute actively while preparing my proposal. I have a quick question: Looking forward to your guidance. Thanks again! |
Beta Was this translation helpful? Give feedback.
-
|
Hi @bdmorgan, @gsquared94 — I'm Sundram ([@SUNDRAM07](https://github.com/SUNDRAM07)), and I've submitted my GSoC 2026 proposal for Project 1: Terminal-Integrated Performance & Memory Investigation Companion ([#23365](#23365)). I've been contributing to Gemini CLI since February with 13 PRs across core, CLI, policy, and MCP subsystems — including a security fix where a typo in policy TOML parsing silently bypassed tool validation ([#20699](#20699)). For my GSoC proposal, I've built the complete prototype:
The prototype delivers all 6 expected outcomes from the issue — automated 3-snapshot technique, heap dump analysis, Perfetto trace integration, CDP-based debugger skills, CPU/memory profiling, and LLM-driven root cause analysis with 9 pattern detectors. Key technical achievement: 85,106x token compression — a 302MB production heap snapshot compressed to ~930 structured tokens while preserving retainer chains and root-cause signals. This means the Gemini agent can diagnose multi-GB heap issues within a single context window. I also contributed to the Performance Monitoring discussion ([#20009](#20009)) with a working prototype and 27 unit tests, and submitted Draft PR [#21262](#21262) for that project. Detailed comment with full architecture breakdown on the issue: [#23365 comment](#23365) Looking forward to the review period. — Sundram (SUNDRAM07) |
Beta Was this translation helpful? Give feedback.
-
|
Hi @akh64bit, I’m Hung, and I’ve just submitted my draft proposal for the "Long-Context & Complex Reasoning Coding Evaluation Dataset" project on the GSoC dashboard. My approach focuses on creating a "Needle-in-a-Codebase" framework, specifically testing Gemini's ability to handle cross-file dependencies in large OSS repositories (50k+ tokens). I plan to use LLM-as-a-judge for evaluating complex reasoning steps. I would love to hear your thoughts on whether you'd prefer the dataset to focus on specific languages (e.g., Go/Python) or if a language-agnostic approach is more aligned with the Gemini CLI roadmap. I'm eager to refine my proposal based on your guidance. Thank you! |
Beta Was this translation helpful? Give feedback.
-
|
Hi @gundermanc , I've been reading through the existing evals/ infrastructure and evals/README.md while drafting my proposal for #23331. |
Beta Was this translation helpful? Give feedback.
-
|
Hi! @bdmorgan @akh64bit Features:
Repo: https://github.com/yashcodes25/gemini-long-context-eval Would love your feedback! |
Beta Was this translation helpful? Give feedback.
-
|
Hi @bdmorgan, circling back on my earlier comment with something more concrete, I've been heads-down on Project 1 (Terminal-Integrated Performance & Memory Investigation Companion, #23365) and wanted to share where things landed. I built a working prototype and ran it against gemini-cli v0.36.0-nightly. The engine parses the full 89 MB production heap snapshot in 1.7 seconds using a streaming parser, character-by-character through The piece I'm most interested in pushing further is the retainer chain extraction. Most heap analysis tools tell you what leaked. The retainer chain tells you why it can't be garbage collected — a BFS walk from the leaked object backward through the reverse heap graph to GC roots. That's the difference between "EventEmitter grew by 200 objects" and "EventEmitter is retained through One thing I've been thinking about, once these tools land, they'll need behavioral evals (#23331). And heap analysis is a harder eval target than most tools because the agent isn't just reading a file or running a command, it's interpreting a complex data structure and making a judgment call about root cause. You could construct known leak scenarios (say, an event listener that's never removed) and check whether the agent identifies the correct retention path and suggests the right fix category. The ground truth is known, so the eval has a clear pass/fail signal. Not sure if that's in scope for this project or if it falls under the evals project, but it seemed worth flagging. I've submitted two upstream PRs that touch the telemetry subsystem — #23536 adds edge-case test coverage for Proposal submitted for Project 1. Prototype: github.com/Anjaligarhwal/gemini-cli-perf-companion (301 tests passing). |
Beta Was this translation helpful? Give feedback.
-
|
Hello, I’m interested in contributing to the "Optimize Test Suite Output Logging" project for GSoC 2026. I have experience working with CLI tools, improving error handling, and enhancing logging systems through my contributions to the gemini-cli repository. I’ve also worked on debugging and improving developer experience in open-source projects. I’m currently exploring the codebase and would like to understand the existing logging workflow better. I plan to start contributing by identifying areas where logging can be made more structured and efficient. Please let me know if there are specific areas I should focus on. Thank you! |
Beta Was this translation helpful? Give feedback.
-
|
Hi everyone! I'm Jayaditya G (@JayadityaGit), and I’ve been a core contributor to Gemini CLI for almost a year, focusing on enhancing the UX/DX. I’ve submitted my GSoC 2026 proposal: "Adaptive Discovery: Personalized UX and Automated Knowledge Sync for Gemini CLI." I originally worked to blend educative tips with witty phrases (#10569), along with other contributions like the The Vision:
I’m excited to build a framework that empowers every developer using the CLI! My Profiles:
prototype :- #26048 |
Beta Was this translation helpful? Give feedback.
-
|
Bruh, only two accepted. |
Beta Was this translation helpful? Give feedback.
-
|
welcome to real world |
Beta Was this translation helpful? Give feedback.
-
|
it takes 30 days for them to rule out an idea which they proposed, it looks like gemini cli is not interested in first place and had to propose some last minute features/developments. it's such a shame. by far the worst gsoc I ever had. |
Beta Was this translation helpful? Give feedback.
-
That is not how this works. If you consider the process a waste, it's on you. I personally have 3 p1 merges in this Gemini repo, including a major issue fix that impacted 100s of people, with multiple 200/300 line code diffs. I also submitted all 3 professional proposals to And I wasn't selected either, doesn't mean it's all gone to waste. I loved every second of it. And now I'm a proud open source contributor. I understand the entire org was really disorganized, there was a lack of communication, no reviews from the mentors or maintainers, but it's their first time as an org as well. Could they have done better? For sure. This is the real world, learn from it, and be back again next year. |
Beta Was this translation helpful? Give feedback.
-
|
I dont think even though it's their first time, not considering submissions for the ideas they proposed is the worst one can do. dont say to me that in bigtech requirements and business needs change time to time, if you are not capable just dont join gsoc as an org. |
Beta Was this translation helpful? Give feedback.
-
|
it's ok no problem, the real thing is to have fun contributing, I did not know that this repo is going to participate in gsoc in the first place, but i just love contributing here... the maintainers themselves never done gsoc, yet they are impacting the project profoundly... lastly i want to say from my experience is that :- "The doors might open, when you really have an impact on the repo" good luck people. |
Beta Was this translation helpful? Give feedback.
and others
Uh oh!
There was an error while loading. Please reload this page.
-
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:
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!
Beta Was this translation helpful? Give feedback.
All reactions