Agents IA
rivet
Rivet is Ironclad's visual IDE and TypeScript library for building complex AI agents and prompt chains
★ 4,7 k étoilesMIT
Voir le guide →OmAgent est un projet de agents IA open source, créé par om-ai-lab (om-ai-lab/OmAgent sur GitHub). Il compte 2 667 étoiles sur GitHub et 291 forks, il est publié sous la licence Apache-2.0 et a été mis à jour pour la dernière fois sur GitHub le 19 mars 2025.
Python library for building multimodal language agents — VLM, video and mobile, fast to prototype and ship
Pick a multimodal task — e.g. answering questions about a video, a visual QA web app, or a phone-connected assistant.
pip install omagent-core, or clone the repo to work from the examples.
Generate container.yaml for an example with compile_container.py, and set your LLM key/endpoint (or point it at a local Ollama model) in configs/llms.
Run the simple VQA example's webpage demo and ask a question about an image.
A flexible architecture with graph-based workflow orchestration and various memory types for contextual reasoning.
Built-in support for VLM models, real-time APIs, computer vision models, video processing and mobile device connection.
State-of-the-art unimodal and multimodal agent algorithms beyond simple LLM reasoning, e.g. ReAct, CoT and SC-CoT, packaged as reusable operators.
Deploy your own models locally with Ollama or LocalAI instead of hosted APIs.
A fully distributed architecture with custom scaling, or Lite mode that eliminates middleware deployment.
Video QA agent (Gradio app), mobile personal assistant, and an agentic-operators comparison project are included.
Clone OmAgent to use its examples; the library itself installs from PyPI in step 2.
git clone https://github.com/om-ai-lab/OmAgent.git
cd OmAgentInstall omagent_core from PyPI (Python >= 3.10).
pip install omagent-coreGenerate the example config and launch the webpage demo on http://127.0.0.1:7860.
cd examples/step1_simpleVQA
python compile_container.py
python run_webpage.pyom-ai-lab maintains it; the framework paper is Zhang et al., 2024 (arXiv:2406.16620), presented in the EMNLP 2024 context.
Yes, Apache-2.0 licensed.
Only if you choose a hosted LLM/VLM — set your OpenAI-compatible key in the config, or run local models with Ollama/LocalAI instead.
The authors aim for a simpler interface with less overhead, while natively covering multimodal inputs (video, vision, mobile) that text-first frameworks treat as add-ons.
pip install omagent-core, then run the step1_simpleVQA example: compile_container.py, configure your LLM, and python run_webpage.py.
Rivet is Ironclad's visual IDE and TypeScript library for building complex AI agents and prompt chains
Google's code-first toolkit for building real AI agents.
Mobile-Agent — multimodal GUI agents that operate phones and PCs from screenshots and instructions.
Google's Python SDK for building AI agents on the Antigravity platform and Gemini models.
smolagents — Hugging Face's barebones agent library where agents think in code: a CodeAgent that writes its actions as Python, model-agnostic, Apache-2.0
A framework for building, orchestrating and deploying AI agents and multi-agent workflows with support for Python and .NET.
Les autres projets de la catégorie Agents IA avec le plus d'étoiles du catalogue LibreAI — la sélection réaliste d'alternatives si OmAgent ne vous convient pas.
A personal AI that really does things — on any OS, with your data staying yours.
An agentic skills framework & software development methodology that works.
Agent harness optimization system with skills, memory and security for Claude Code, Codex and more
The agent that grows with you — a Python framework for adaptive AI agents.
DeepSeek's plugin-based agent harness where everything is a plugin.
Projets de la catégorie Agents IA proches dans le catalogue, les plus proches en étoiles de OmAgent — continuez à explorer de part et d'autre.
Free Chinese AIGC course: prompting, agents, image, video, voice, open models.
Self-learning browser infrastructure for AI agents — sites remembered, so agents stop rediscovering the web
YAO: rigorous engineering, evaluation, and governance for reusable agent skills.
Bright Data's MCP server: web search, scraping, structured data and browser automation for agents (75 tools)
Official agent skills that teach AI coding tools to work with Supabase.
NVIDIA's toolkit for connecting, profiling and optimizing teams of AI agents — across any framework
Gratuit et ouvert : récupérez le code source, mettez une étoile au dépôt et commencez à créer.