GSoC idea #12 Demo PoC and Open-Discussion | Terminal-native visualization pipeline integrated into Gemini CLI

[Aaxhirrr]

PoC Demo Submission (Open for Discussion)

Hi everyone, aashir here,

I wanted to share my Proof-of-Concept demo for terminal visualization in Gemini CLI (GSoC Idea 12 direction).
This is not intended as a merge-ready final patch yet. I’m sharing it early so I can get feedback on approach, scope, and architecture.

I integrated the work directly into the Gemini CLI codebase (not as a separate app), to demonstrate the understanding of the codebase, and focused on showing that this can work end-to-end in real terminal workflows.

I know this is far from perfect right now.
There are still rough edges, and I’m treating this as a practical learning + validation pass, not a finished implementation.

I’ve already contributed to gemini-cli, and gotten two prs approved, and one merged. I'd really love the chance to officially work on this, iterate with maintainer feedback, and help push it toward a production-quality state.

Video Demo

GSoC_POC12.mp4

What I’m Demonstrating

• Inline diagram rendering in graphics-capable terminals
• ASCII fallback for non-graphics/basic terminals
• /visualize flows from prompt/file
• /graph git flow
• /explain --visualize flow
• /preview flow for UI rendering

Links

• Draft PR: [PoC][GSoC #12] Generative Architecture & UI Visualization (termviz integration) #22072
• PoC branch: https://github.com/Aaxhirrr/gemini-cli/tree/feature/termviz-integration
• Demo guide: https://github.com/Aaxhirrr/gemini-cli/blob/feature/termviz-integration/DEMO.md
• Video Link: https://docs.google.com/videos/d/1jXJkx9n7V8yOyfcV54AHZuWUJIG9NEY2DxYkFKk3dSI/edit?scene=id.p#scene=id.p

Important Note

I know the diff is large, but this is a proposal-style PoC branch and includes supporting visualization plumbing needed for runnable demos. I’m very open to splitting/restructuring this into a cleaner stacked path based on maintainer guidance.

Quick Intro

• I’m Aashir Javed, a Computer Science (ML track) student at Arizona State University (GPA: 4.0, graduating Dec 2026).
• I’m currently a Machine Learning Research Intern (FURI), working on memory-aware neuro-symbolic + graph-based systems for early Alzheimer’s progression modeling.
• At Sedai (Generative AI Division), I built AWS Bedrock-powered log-analysis tooling across 5+ microservices, reducing triage time by ~35% and parsing errors by ~40%.
• I’m also leading autonomy software for a solar-powered beach-cleanup robot project in collaboration with the Polytechnic University of Barcelona.
• I’ve contributed as a Beta Contributor at CreateAI Lab, evaluating and stress-testing 30+ LLMs with faculty stakeholders.
• I've won 5x hackathons including Nvidia and Anthropic.
• Core stack: Python, TypeScript, React/Node/FastAPI, AWS/GCP, Docker, Postgres/Neo4j, LLM systems.

Excited for any and every feedback, from @LyalinDotCom @bdmorgan and all the passionate contributors out there. Feel free to start an open discussion about the idea.

Thanks for taking a look, and for your time.