#shell #bonsai #llamacpp #llm #mlx #prism_ml #small_models

You can run powerful Bonsai 27B AI models locally on your iPhone, Mac, or laptop without needing cloud servers. The 1-bit version fits in just 3.9 GB, allowing a 27-billion-parameter model (with vision and reasoning) to run on a phone like the iPhone 17 Pro, while the Ternary version offers higher quality at 5.9 GB. This gives you private, fast, and offline AI for chat, image analysis, and tool use, with up to 262K token context and 90–95% of full-precision performance.

https://github.com/PrismML-Eng/Bonsai-demo

#python #ai_coding #claude #claude_code #code_review #graphrag #incremental #knowledge_graph #llm #mcp #python #static_analysis #tree_sitter

code-review-graph is a free, local-first tool that builds a structural map of your code to stop AI assistants from re-reading your entire codebase, saving you significant money and time. By using Tree-sitter to parse your code into a graph and tracking changes incrementally, it identifies the exact "blast radius" of a change so your AI only reads the few files that actually matter. This reduces token usage by an average of 82x (up to 528x in best cases), cutting costs and making AI reviews feasible for massive projects like monorepos.

https://github.com/tirth8205/code-review-graph

#typescript

This repository offers free, hands-on workshop materials from Anthropic’s Code with Claude program to teach you how to build advanced AI agents using Claude Managed Agents. You will learn to create multi-agent systems, add memory for cross-session persistence, optimize model costs, and verify agent performance with automated evaluations. The benefit is that you gain practical skills to ship production-ready AI agents efficiently, moving from simple scripts to complex workflows that handle real tasks like log analysis, research, and incident dashboards without needing to guess if your code works.

https://github.com/anthropics/cwc-workshops

#typescript

G0DM0D3 is a free, open-source chat tool that lets you run 50+ AI models (like GPT, Claude, and Grok) in your browser at once to find the best, unfiltered response . You get full privacy because all your data and history stay in your browser, with no accounts or server storage . The main benefit is control: you can test models side-by-side, bypass safety filters using tools like "GODMODE CLASSIC," and tune settings automatically without paying monthly fees . It’s ideal for researchers, hackers, and anyone wanting unrestricted AI access .

https://github.com/elder-plinius/G0DM0D3

#typescript #agent #ai #ai_agent #claude #cli #developer_tools #local_first #mcp #mcp_server #metasearch #model_context_protocol #nodejs #privacy #rag #search #search_engine #typescript #web_crawler #web_scraping #web_search

wigolo is a free, local-first tool that gives your AI agent web search, crawling, fetching, and research capabilities without needing API keys, cloud services, or paying per-query bills . It runs on your machine as an MCP server, keeping all data private and stored locally . The benefit to you is zero cost per query and full privacy, while still delivering web intelligence quality that rivals paid services . You can set it up with one command and immediately use powerful tools like autonomous research and page monitoring without worrying about metered costs .

https://github.com/KnockOutEZ/wigolo

#python

LingBot-Map is an open-source AI model that instantly builds accurate 3D maps from regular video streams at 20 frames per second, handling over 10,000 frames without slowing down . It uses a special "Geometric Context Transformer" to fix camera drift and keep geometry stable as you move . You benefit by getting real-time 3D understanding for robots, autonomous vehicles, or AR apps using only a standard camera, eliminating the need for expensive depth sensors or slow offline processing .

https://github.com/Robbyant/lingbot-map

#python

KTransformers lets you run and fine-tune massive 100B+ AI models on your single consumer GPU (like an RTX 4090) by smartly using your computer's CPU memory and power alongside the GPU. This saves you money by avoiding expensive multi-GPU server clusters while still delivering high speed, allowing you to use cutting-edge AI locally without sacrificing performance.

https://github.com/kvcache-ai/ktransformers

#rust

Jellium Desktop is an unofficial, fast desktop app for watching your Jellyfin media on Linux, macOS, and Windows. It uses modern video tech (CEF and mpv) to improve performance compared to older versions. You benefit by getting a reliable, high-quality streaming experience with easy downloads for your specific computer type, including Apple Silicon and Intel macOS, plus Windows and Linux options, without needing to pay or use official support.

https://github.com/andrewrabert/jellium-desktop

#python #agent #agent_memory #ai_agent #book #coding_agent #context_engineering #large_language_models #llm #mcp #multi_agent #multimodal #rag #reinforcement_learning

This free open-source book and code library teaches you how to build AI Agents using the simple formula: Agent = Model + Context + Tools. You get the full book in English, Chinese, and Tamil, plus ready-to-run code examples for every chapter. The benefit is that you can learn by doing instead of just reading, letting you build real AI agents that search the web, write code, remember users, and work with other agents to solve complex tasks.

https://github.com/bojieli/ai-agent-book

#typescript #a2a #ai_agents #ai_gateway #anthropic #claude #claude_code #cline #codex #copilot #cursor #deepseek #free_ai #gemini #kimi #llm_gateway #mcp #openai #openai_proxy #qwen #token_saver

OmniRoute is a free tool that connects your AI coding apps to 268 providers, including 90+ with free tiers, so you can use top models like Claude or GPT without paying. It automatically switches to another provider if one hits its limit and compresses your requests to cut token usage by up to 95%. You benefit by saving money, avoiding interruptions, and getting unlimited access to powerful AI tools through a single, easy setup.

https://github.com/diegosouzapw/OmniRoute

#typescript

Ontology Playground is a free, open-source web app that lets you learn about ontologies and Microsoft Fabric IQ by exploring pre-built models or designing your own with a visual editor. You can export your work as RDF/XML, share interactive diagrams, and access 9 structured courses with labs—all without needing a backend. This helps you build and test knowledge graphs safely before implementing them in real systems, saving time and reducing errors in data architecture.

https://github.com/microsoft/Ontology-Playground

#typescript #backlink_analysis #google_search_console_mcp #keyword_research #mcp #seo #seo_tools #site_audit

OpenSEO is a free, open-source SEO tool that replaces expensive platforms like Semrush with a simple, pay-as-you-go model where you only pay for the data you use. It connects directly to AI agents like Claude via MCP, allowing your AI to research keywords, track ranks, and analyze backlinks using real data instead of guesses. This saves you money by eliminating monthly subscriptions and boosts efficiency by automating tedious SEO tasks while giving you full control over your tool.

https://github.com/every-app/open-seo

#typescript #agents #ai #deployments #self_hosted

Openship is an open-source, self-hostable deployment platform with built-in CI/CD that lets you push code, build containers, and manage infrastructure like databases, SSL, and backups from a desktop app, web dashboard, or CLI . You benefit by shipping apps in under two minutes with zero config files or YAML pipelines, keeping your data private on your own servers while avoiding traditional CI/CD complexity and vendor lock-in .

https://github.com/oblien/openship

#rust

Topcoat is an experimental, full-stack Rust framework that lets you build modern web apps with instant client interactivity without WebAssembly or separate build steps. It renders all markup on the server, so your components can directly query databases and skip the traditional API layer, while `$()` expressions run as type-checked Rust on the server and translate to JavaScript for instant browser updates. You get module-based routing, built-in Tailwind CSS, editable UI components, and asset bundling. The benefit is faster development with less boilerplate, simpler code, and high performance leveraging Rust’s safety and Tokio’s concurrency.

https://github.com/tokio-rs/topcoat

#cplusplus #asr #ggml #gguf #speech_to_text

transcribe.cpp is a fast, free C/C++ library that turns speech into text using 16 model families (like Whisper and Parakeet) on your own GPU or CPU, with official supports for Python, JavaScript, Rust, and Swift. You benefit by getting accurate, verified transcriptions without paying cloud API fees, running entirely on your device for privacy, and supporting streaming, batch, and specialized tasks like medical dictation or speaker diarization.

https://github.com/handy-computer/transcribe.cpp

#python #cfg #generative_ai #json #llms #prompt_engineering #regex #structured_generation #symbolic_ai

Outlines is a Python library that guarantees large language models produce structured outputs (like JSON, XML, or custom schemas) directly during generation, not after. By simply specifying your desired output type using Python syntax (e.g., `int` or a Pydantic model), it mathematically ensures the result matches your structure exactly. This benefits you by eliminating broken JSON, parsing errors, and fragile regex code, making your AI applications reliable, predictable, and ready for production use without post-generation fixes.

https://github.com/dottxt-ai/outlines

#rust #gguf #llm #localai #mlx #skill #unsloth

llmfit is a free terminal tool that automatically checks your computer’s RAM, CPU, and GPU to tell you exactly which large language models will run well on your machine, ranking them by fit, speed, quality, and context. It supports both an interactive visual interface and a command-line mode, works with multi-GPU setups, and suggests the best quantization levels so you never waste time or storage downloading models that won’t work. This saves you from guesswork, trial-and-error testing, and wasted bandwidth, giving you immediate, reliable recommendations for the best local AI models for your specific hardware.

https://github.com/AlexsJones/llmfit

#typescript

pi-web is a browser interface for the pi coding agent that lets you browse past sessions, chat in real time, manage models and skills, and preview project files without using the terminal . You get clearer results with structured tool calls, easy session branching to test different ideas safely, and quick access to previous conversations by project . This saves time by avoiding terminal history digging and makes complex agent work more visual and manageable.

https://github.com/agegr/pi-web

#javascript

The TradingView MCP Bridge lets your AI assistant (like Claude) see and control your local TradingView charts, enabling AI-powered chart analysis, Pine Script coding, and workflow automation without sending data off your machine. You benefit by writing and debugging trading scripts, reading indicator values, drawing lines, and navigating charts using simple text commands instead of clicking menus, saving time and boosting your trading research efficiency—all while keeping your data secure locally and requiring no paywall bypass.

https://github.com/tradesdontlie/tradingview-mcp

#javascript #3mf #agents #ai_agents #build123d #cad #dxf #glb #mechanical_engineering #opencascade #robotics #sdf #srdf #step #stl #stp #text_to_cad #urdf

CAD Skills is a free, open-source library that lets AI agents like Codex and Claude Code create, edit, and verify CAD models, robot files, and 3D print code from simple text or images . You benefit by automating complex design tasks—such as generating STEP parts, writing robot descriptions (URDF/SRDF), slicing G-code, and previewing files locally—without manual modeling, saving time and reducing errors in hardware and robotics projects .

https://github.com/earthtojake/text-to-cad

#other #adhd #claude_ #claude_code_plugin #claude_skills #developer_tools #productivity

This plugin makes your coding assistant give ADHD-friendly answers by starting with the next action, using numbered steps, and cutting out fluff like "Hope this helps!" No diagnosis is needed. You install it for Claude Code or Codex with simple commands, then type `/i-have-adhd` or `$i-have-adhd`. The benefit is that you get clear, direct steps immediately, saving you from scrolling past long explanations and helping you stay focused and finish tasks faster.

https://github.com/ayghri/i-have-adhd