$ show Shubhamsaboo/awesome-llm-apps
This repository is a vast curated collection of over 100 open-source AI agent application templates and tutorials, spanning multiple frameworks (Agno, Google ADK, OpenAI SDK, LangGraph, CopilotKit, CrewAI) and deployment patterns (Streamlit, Next.js, FastAPI, CLI). It serves as a living portfolio of AI agent patterns: single-agent assistants, multi-agent teams, generative UI (CopilotKit + AG-UI), RAG pipelines with various vector stores (Qdrant, Chroma, LanceDB, FAISS), voice agents, always-on scheduled agents, and agent skills (code analysis, dependency health). The architecture is a monorepo of independent sub-projects, each with its own requirements, entry point, and often a README. External dependencies are heavy: nearly every app requires API keys for LLMs (OpenAI, Google, Groq, Together) and tools (Tavily, Firecrawl, SerpAPI, E2B), plus optional services (PostgreSQL, Qdrant, ElevenLabs). While the codebase is not a single integrated application, it provides a rich set of reference implementations for developers building AI agents.
The LLM identified 5 files worth highlighting for reuse:
A standalone Python script that reconstructs file and line history from local git logs, classifies commits into categories (feature, fix, refactor), detects co-change patterns and intent signals (revert, workaround, TODO). It is decoupled from any framework, works purely via git CLI, and outputs structured JSON. This is reusable for any project that needs automated commit classification or historical code analysis.
An Agno agent that generates podcast scripts as structured Pydantic models (PodcastScript) using USE_JSON_MODE. It takes scraped content and produces a full script with host dialogue, segments, and source citations. The pattern of wrapping LLM output in a validated schema is reusable for any content generation pipeline (social posts, reports, tutorials).
Encapsulates Tic-Tac-Toe game logic in a clean, stateless TicTacToeBoard class with move validation, winner detection, and board representation. It is fully isolated from the agent or UI layers, making it reusable for any Python game (e.g., as a tool for LLM agents, or integrated into different frontends). Demonstrates good separation of concerns.
A Google ADK LlmAgent that dynamically generates a dashboard UI by utilizing tools to update state (add/remove charts, set themes, toggle filters). It uses before_model_callback to inject current dashboard state and after_model_callback to log actions. This pattern of an agent that manipulates a structured UI state via tool calls is reusable for any generative UI application.
A deterministic evaluation harness that scores agent skill descriptions against trigger case files using lexical matching and quality heuristics. It is framework-agnostic, runs offline, and produces structured JSON results. Reusable for any project that needs to evaluate text-based skill definitions against expected triggers.