#python #agent #agentic_ai #agentic_framework #agentic_workflow #ai #ai_agents #ai_companion #ai_roleplay #benchmark #framework #llm #mcp #memory #open_source #python #sandbox

MemU lets AI systems take in conversations, documents, and media, turn them into structured memories, and store them in a clear three-layer file system. It offers both fast embedding search and deeper LLM-based retrieval, works with many data types, and supports cloud or self-hosted setups with simple APIs. This helps you build AI agents that truly remember past interactions, retrieve the right context when needed, and improve over time, making your applications more accurate, personal, and efficient.

https://github.com/NevaMind-AI/memU

#javascript #agent #agentic #agentic_ai #ai #ai_agents #automation #cursor #design #figma #generative_ai #llm #llms #mcp #model_context_protocol

Cursor Talk to Figma MCP lets Cursor AI read and edit your Figma designs directly, using tools like `get_selection` for info, `set_text_content` for bulk text changes, `create_rectangle` for shapes, and `set_instance_overrides` for components. Setup is quick: install Bun, run `bun setup` and `bun socket`, add the Figma plugin. This saves you hours by skipping context switches, automating repetitive tasks like text replacement or override propagation, speeding up design-to-code workflows, and keeping everything in sync for faster, precise builds.

https://github.com/grab/cursor-talk-to-figma-mcp

#typescript #acp #ai #ai_agent #banana #chat #chatbot #claude_code #codex #cowork #excel #gemini #gemini_cli #gemini_pro #llm #multi_agent #nano_banana #office #qwen_code #skills #webui

AionUi is a free, open-source app that gives your CLI AI tools like Gemini CLI, Claude Code, and Qwen Code a simple graphical interface on macOS, Windows, or Linux. It auto-detects them for easy chatting, saves talks locally with multi-sessions, organizes files smartly, previews 9+ formats like PDF or code instantly, generates/editing images, and offers web access. You benefit by ditching complex commands for quick, secure AI help in office tasks, coding, or data work—saving time and boosting productivity without data leaving your device.

https://github.com/iOfficeAI/AionUi

#jupyter_notebook #chinese_llm #chinese_nlp #finetune #generative_ai #instruct_gpt #instruction_set #llama #llm #lora #open_models #open_source #open_source_models #qlora

AirLLM is a tool that lets you run very large AI models on computers with limited memory by using a smart layer-by-layer loading technique instead of traditional compression methods. You can run a 70-billion-parameter model on just 4GB of GPU memory, or even a 405-billion-parameter model on 8GB, without losing model quality. The benefit is that you can use powerful AI models on affordable hardware without expensive upgrades, and the tool also offers optional compression features that can speed up performance by up to 3 times while maintaining accuracy.

https://github.com/lyogavin/airllm

#python #deepseek #demo #easy #embedding #flask #gpt #huggingface_transformers #llm #mcp #multimodal #openai #qwen #rag #sentence_transformers #ui #vllm #vlm

UltraRAG is a lightweight framework that makes building retrieval-augmented generation (RAG) systems simple and fast. It uses a low-code approach where you write just dozens of lines of YAML configuration instead of complex code to create sophisticated AI workflows with conditional logic and loops. The framework includes a visual development environment where you can drag-and-drop to build pipelines, adjust parameters in real-time, and instantly convert your logic into interactive chat applications. This means you can deploy powerful AI systems that ground answers in your own data—reducing hallucinations and improving accuracy—without needing extensive coding expertise or lengthy development cycles.

https://github.com/OpenBMB/UltraRAG

#python #abliteration #llm #transformer

Heretic is an automated tool that removes safety restrictions from AI language models while preserving their intelligence and capabilities. It uses advanced mathematical techniques called directional ablation to identify and disable the "refusal mechanisms" that prevent models from answering certain questions. The key benefit is that anyone can use it with a simple command—no technical expertise needed. Unlike manual methods that often damage model quality, Heretic achieves the same level of censorship removal with significantly better preservation of the original model's reasoning abilities, as measured by lower KL divergence scores. This means you get an uncensored model that still thinks clearly and produces high-quality responses.

https://github.com/p-e-w/heretic

#python #ai #claude #gemini #llama #llm #openai

You can access powerful AI language models for free or with trial credits through multiple legitimate platforms. Services like OpenRouter, Google AI Studio, Groq, and Mistral offer free tiers with varying request limits, while others like Fireworks, Baseten, and Inference.net provide trial credits ranging from $1 to $30. These platforms support diverse models including Llama, Gemma, Qwen, and DeepSeek, enabling you to build and test AI applications without upfront costs. The benefit is clear: you can prototype, develop, and deploy AI-powered features while managing your budget effectively, with options to scale up as your needs grow.

https://github.com/cheahjs/free-llm-api-resources

#rust #ai #ai_ocr #attention_mechanism #gnn #gnn_model #gnns #graph #graph_neural_networks #llm_inference #low_latency #mincut #neo4j #ocr #onnx #rust #vector #wasm

RuVector is a free, open-source vector database that gets smarter with every query. Unlike static databases, it learns from usage via GNN layers, runs LLMs locally with no cloud costs, supports graph queries like Neo4j, scales freely across nodes, and deploys as a single self-booting file (125ms startup). Run with `npx ruvector`. You benefit from faster, more accurate AI search that improves automatically, zero operating costs, full offline/privacy control, and easy scaling—perfect for RAG, agents, or edge apps without vendor lock-in.

https://github.com/ruvnet/ruvector

#rust #ai_gateway #ai_gateway_support #envoy #envoyproxy #gateway #generative_ai #llm_gateway #llm_inference #llm_proxy #llm_routing #llmops #llms #openai #prompt #proxy #proxy_server #routing

Plano is an AI-native proxy server that handles key tasks for agentic apps like routing between agents, smart LLM model selection, safety guardrails, and automatic traces for observability. Define agents in simple YAML, write basic HTTP code in any language, and start Plano to run multi-agent systems without custom plumbing or framework lock-in. You benefit by building and shipping reliable agents to production much faster, focusing on core logic while gaining safety, low latency, and easy scaling.

https://github.com/katanemo/plano

#javascript #ai #algorithm #artificial_intelligence #chatgpt #claude #cursor #deep_learning #deepseek #gemini #generative_ai #gpt #llm #mcp #openai #python #rag #vibe_coding #vibecoding #vue #vuepress

鱼皮的 AI知识库 offers a free Vibe Coding tutorial for beginners, teaching AI-powered programming with natural language prompts to build and monetize apps fast—no coding skills needed. It covers tools, projects, tips, and paths like making your first work in 10 minutes, plus AI guides on DeepSeek, Cursor, and more. You benefit by quickly creating profitable products, breaking tech barriers, and enjoying AI perks to improve life and work. Start at ai.codefather.cn/vibe.

https://github.com/liyupi/ai-guide

#typescript #agent #agentic #agentic_framework #agentic_workflow #ai #ai_agents #bytedance #deep_research #harness #langchain #langgraph #langmanus #llm #multi_agent #nodejs #podcast #python #superagent #typescript

DeerFlow 2.0 is an open-source super agent harness that orchestrates multiple sub-agents, memory systems, and sandboxed execution environments to accomplish complex tasks. Built on LangGraph and LangChain, it combines research, coding, and content creation capabilities with extensible skills and tools. The platform features isolated Docker containers for safe execution, long-term memory that learns your preferences, and the ability to spawn sub-agents that work in parallel on different task angles. You benefit from dramatically reduced research and automation time—tasks that typically take hours complete in minutes—while maintaining full transparency and control over agent decisions through human-in-the-loop collaboration. Whether you need deep research reports, data analysis, slide decks, or custom workflows, DeerFlow handles multi-step complexity without requiring extensive coding knowledge.

https://github.com/bytedance/deer-flow

#typescript #agent #agent_development #ai_agent #claude #claude_code #educational #llm #python #teaching #tutorial

Claude Code is an AI agent framework that uses a simple loop: send messages to Claude, check if it needs tools, execute those tools, and repeat. The benefit is that you can build powerful autonomous agents by layering one feature at a time—from basic tool use to multi-agent teams—without rewriting the core loop. This modular approach lets you start simple with bash commands and scale to complex workflows with planning, skill loading, background tasks, and team coordination, making it easier to automate development work and delegate entire projects to AI agents.

https://github.com/shareAI-lab/learn-claude-code

#typescript #agentic_ai #ai_agents #claude_code #cli #codex #coding_agents #cursor_agent #desktop_app #developer_tools #electron #git_worktree #llm #mcp #opencode #orchestration #parallel_agents #terminal #tui #vibe_coding #worktrees

Superset is a turbocharged macOS terminal for running 10+ CLI coding agents like Claude Code, Cursor, and GitHub Copilot in parallel. It isolates tasks in separate Git worktrees to avoid interference, lets you monitor progress from one dashboard, review changes with a built-in diff viewer, and switch contexts quickly. You benefit by coding 10x faster, shipping more without context-switching delays or conflicts, saving time on development workflows.

https://github.com/superset-sh/superset

#python #agent #chatbot #large_language_models #llm #llm_agent #mcp #multi_agent #multi_modal #react_agent

AgentScope is a simple, production-ready framework to build AI agents fast. Install with `pip install agentscope` (Python 3.10+), then create ReAct agents with tools, memory, voice, human steering, multi-agent workflows, and finetuning in 5 minutes. It supports realtime voice, A2A protocols, RL training, and easy deployment locally, in cloud, or Kubernetes. You benefit by quickly making robust, scalable agents for tasks like games, research, or chats without complex coding, saving time and enabling real-world apps.

https://github.com/agentscope-ai/agentscope

#python #agent #llm #llm_agent #llm_reasoning #machine_learning_systems #mlsys #reinforcement_learning #rl

AReaL is a free, open-source system for fast asynchronous reinforcement learning to train large AI models in math, coding, search, and agents. It decouples generation and training for up to 2.77x speedup, stable performance, and easy setup on single or 1000+ GPUs with algorithms like GRPO/PPO. Install via git/pip, run examples like GSM8K math instantly. You benefit by building top AI agents affordably and quickly, reproducing results with shared data/models, saving time/money vs. slow synchronous tools.

https://github.com/inclusionAI/AReaL

#python #agent #agentic_rag #ai_agents #clawbot #context_database #context_engineering #filesystem #llm #memory #openclaw #opencode #rag #skill

OpenViking is a free open-source tool that acts as a context database for AI agents, using a simple file system to organize memories, resources, and skills under viking:// paths. It fixes issues like scattered data, high token costs, weak searches, and untraceable errors with tiered loading (L0 abstracts, L1 overviews, L2 details loaded on demand), recursive directory retrieval, visual traces, and auto-session memory updates. You benefit by building smarter, cheaper agents faster—like managing files—saving up to 96% on tokens while boosting task success by 50%+.

https://github.com/volcengine/OpenViking

#python #agent #agents #ai #anthropic #claudecode #llm #llms #openai

Open SWE is a free, open-source framework to build internal coding agents like those at Stripe, Ramp, and Coinbase. Trigger it via Slack, Linear, or GitHub (@openswe) to research codebases, plan tasks, code, test, review, and auto-open PRs in secure cloud sandboxes—running parallel jobs without your machine's resources. Customize models, tools, and workflows easily. You benefit by automating routine coding, slashing review cycles and production time by 30-50%, freeing you to focus on high-value work while ensuring safe, high-quality changes.

https://github.com/langchain-ai/open-swe

#python #agent #finance #llm #multiagent #trading

TradingAgents is a free, open-source framework using AI agents like analysts, researchers, traders, and risk managers to mimic real trading firms and make smart stock decisions. Install easily via GitHub, set API keys for models like GPT or Claude, and run CLI or Python code for quick analysis on any ticker. It boosts returns up to 30.5% yearly with low risk and clear explanations. You gain powerful, customizable tools to test and improve trading strategies fast, saving time and spotting better opportunities.

https://github.com/TauricResearch/TradingAgents

#typescript #accounting #ai_analysis #currency_exchange #expenses #invoices #llm #llm_apps #self_hosted #taxes

TaxHacker is a self-hosted AI app that automates accounting for freelancers and small businesses. Upload receipt photos or invoice PDFs, and it extracts key data like amounts, dates, merchants, taxes, and items, storing them in an easy database with auto-categorization, multi-currency conversion (even crypto), custom fields, and exports. You benefit by saving hours on manual entry, simplifying tax reports, ensuring privacy on your server, and focusing on your work instead of paperwork.

https://github.com/vas3k/TaxHacker

#javascript #ai_agents #anthropic #claude #claude_code #developer_tools #llm #mcp #productivity

Everything Claude Code is a top-rated (50K+ stars) plugin system that boosts AI coding agents like Claude Code with 28 agents, 116 skills, 59 commands, hooks, rules, and security scans across 10+ languages (TypeScript, Python, Go, Java, etc.). Install via `/plugin install` for quick setup, then use `/plan`, `/tdd`, or `/code-review` to automate planning, testing, reviews, and builds. It cuts token costs 60%+, enables continuous learning from sessions, and works on Cursor, Codex, OpenCode. You save hours on tedious tasks, get production-ready code faster with less errors, and scale complex projects effortlessly.

https://github.com/affaan-m/everything-claude-code