Shelly

Run OpenAI Codex CLI natively on Android

Vercel DayDeveloper ToolsArtificial IntelligenceGitHubOpen Source

▲ 60 votes1 commentsLaunched Jun 16, 2026

Visit Website

Daily #35Weekly #98

Shelly runs the OpenAI Codex CLI natively on Android — no PC, no Termux, no proot. It pairs a real native terminal with an Agent Chat pane that reads Codex output live and runs fixes with one tap. A home-screen Scouter widget shows quota, cost and rate limits, and lets you fire prompts straight from the home screen. Local LLMs (Qwen3.5 + llama.cpp) supported. Open source, GPLv3. Built entirely by directing AI agents — I don't write the code myself.

AI Analysis

📝 Summary

Shelly enables running OpenAI Codex CLI natively on Android without PC, Termux, or proot. Core features include a native terminal paired with an Agent Chat pane that reads Codex output live and executes one-tap fixes, plus a home-screen Scouter widget for monitoring quota, cost, rate limits and launching prompts instantly. It supports local LLMs via Qwen3.5 and llama.cpp, is fully open source under GPLv3, and was built entirely by directing AI agents. It solves key pain points for mobile developers: complex setup barriers, lack of seamless AI coding on Android, and difficulty tracking API usage on the go. The value proposition is a portable, efficient, AI-powered coding companion that brings powerful dev tools directly to Android home screens.

📈 Market Timing

In 2025-2026, AI agent adoption, local LLM efficiency gains, and mobile-first developer workflows are accelerating rapidly. Demand for on-device AI coding tools is rising with remote work trends and API cost awareness, while open-source AI projects thrive. Shelly's native Android focus, local model support, and AI-built nature align perfectly with these shifts. Excellent Timing.

✅ Feasibility

Technical difficulty is moderate: leverages mature llama.cpp but requires sophisticated native Android terminal and real-time agent integration. Development costs are very low since the product was built entirely by directing AI agents rather than traditional coding. No supply chain issues as pure software; open-source GPLv3 model aids community-driven scalability and maintenance. Compliance risks minimal. High overall feasibility with strong scalability potential.

🎯 Target Market

Primary segments: Android-using software developers, indie hackers, AI tinkerers and open-source enthusiasts, ages 20-40. Industries: software engineering and AI development. Geographic distribution: global with strong adoption in high-Android regions (Asia, US, Europe). Core pain points are cumbersome mobile coding setups and real-time AI assistance gaps. Market is niche yet high-engagement within the growing mobile AI dev tools space. Willingness to pay is moderate (likely via donations, sponsorships or future premium tiers) as base product is open-source.

⚔️ Competition

Low. Direct competitors: 1. Termux (termux.dev) - general Android terminal emulator requiring extra setup. 2. Open Interpreter (openinterpreter.com) - LLM-powered code execution, primarily non-mobile. 3. Aider (aider.chat) - AI pair programming CLI, desktop-focused. 4. Replit mobile app (replit.com) - cloud-based coding environment. 5. Various llama.cpp Android ports (GitHub projects). Advantages: true native execution without proot/Termux, unique one-tap Agent Chat fixes, convenient home-screen widget, local LLM ease, and novel AI-built development story. Disadvantages: Android-only, potential mobile hardware performance limits vs desktop tools, and reliance on user-provided API keys.

Upgrade Pro to unlock full AI analysis