awesome-llm-apps: 100+ AI agent templates you can run today

The usual trap when learning agent development: tutorials written against stale APIs, examples that won’t run, and a month of scaffolding before anything works. Shubhamsaboo’s awesome-llm-apps puts 100+ AI agents, agent skills, and RAG app templates in one repo, with the README’s own promise: “Hand-built, tested end-to-end.” Clone it, add one API key, and it runs — with Claude, Gemini, GPT, DeepSeek, Llama, or Qwen behind it. A Chinese WeChat Channels video called it a 140k-star collection; the actual count as of 2026-10-01 is 140,396, so no exaggeration there.

Core features

• Templates across the stack: single-file starter agents, multi-agent teams, voice agents, and agentic RAG pipelines, organized into browsable categories.
• Drop-in agent skills: packages under agent_skills/ install into coding agents like Claude Code or Codex with one npx skills add command — see self-improving-agent-skills, which rewrites itself against its own evals.
• Real business examples: voice insurance-claim handling, AI home-renovation renders, fraud investigation, an always-on Hacker News briefing agent — worked apps, not hello worlds.
• Backend-agnostic: swap providers by changing environment variables across Claude, GPT, Gemini, DeepSeek, Qwen, and local models.
• Weekly cadence: the README promises new templates weekly, and the repo has shipped steadily since April 2024.

Typical use cases

• Learning agents: start with a single-file template, get one running, then modify — skip the framework scaffolding.
• Product prototypes: adapt the insurance or renovation templates into a working demo; it lands better than slides.
• Augmenting coding agents: hand npx skills add <repo subdir> to Claude Code and it gains the new capability directly.

Quick start

Fastest path — install an existing skill into your agent:

npx skills add https://github.com/Shubhamsaboo/awesome-llm-apps/tree/main/agent_skills/project-graveyard

Or run a template app:

git clone https://github.com/Shubhamsaboo/awesome-llm-apps.git
cd awesome-llm-apps/starter_ai_agents/ai_travel_agent
pip install -r requirements.txt
streamlit run travel_agent.py

Summary

awesome-llm-apps suits developers and product people who want working agent examples and business ideas fast; if you need an orchestration engine rather than templates, a framework like Google ADK is the better stop. Apache-2.0, Python, free to clone and sell. Two caveats: templates call model APIs, so have your own keys and budget ready, and some rely on services that may be blocked in some regions; and with 100+ templates, quality varies by directory — read the README of each before running.