GitHub stars surged for versatile AI agent skills from anthropics/skills and Semantica’s context-aware AI infrastructure. Lightweight foundation models like cactus-compute/needle highlight demand for on-device AI. Security and OSINT tools including megadose/holehe and smicallef/spiderfoot also gained notable attention. Utility projects such as macro-inc/macro’s team workspace and NVIDIA-NeMo/Switchyard’s flexible LLM routing reflect a blend of productivity and AI innovation.
cathrynlavery/diagram-design
⭐ 15,288 stars · +4475 today · HTML
Provides 29 self-contained HTML and SVG editorial diagram types designed for clarity and simplicity, avoiding complex dependencies like Mermaid.
semantica-agi/semantica
⭐ 6,916 stars · +713 today · Python · topics: agent-memory, ai, ai-governance, ai-infrastructure, artificial-intelligence, context-engineering, context-graphs, data-engineering, decision-intelligence, developer-tools, explainable-ai, generative-ai, graph-rag, knowledge-graph, llm, ontology, provenance, python, reasoning, semantic-search
A graph-native platform focused on context and accountability in AI systems, enabling better reasoning and explainability for AI agents.
anthropics/skills
⭐ 169,167 stars · +312 today · Python · topics: agent-skills
Open repository of reusable AI agent skills that power interactive AI agents, drawing broad interest for agent-based AI workflows.
cactus-compute/needle
⭐ 5,082 stars · +769 today · Python · topics: cactus, gemini, gemma, llm, on-device-ai
A small (14MB) foundational language model optimized for running on tiny devices like phones and wearables, enabling local AI inference.
altic-dev/FluidVoice
⭐ 9,936 stars · +76 today · Swift · topics: ai, dictation, ios, llama-cpp, macos, swift
A macOS dictation app with on-device speech-to-text and custom AI enhancements, lauded for speed and privacy-friendly local processing.
unslothai/unsloth
⭐ 71,157 stars · +328 today · Python · topics: agent, chatgpt, deepseek, fine-tuning, gemma, gemma3, gpt-oss, image-generation, llama, llm, llms, openai, qwen, reinforcement-learning, self-hosted, stable-diffusion, text-to-speech, tts, ui, unsloth
Local user interface supporting training and running of various large language and diffusion models, including popular open-source LLMs and image generators.
macro-inc/macro
⭐ 2,696 stars · +1239 today · Rust · topics: agent, ai, ai-agents, all-in-one, crm, crm-system, email, linear, mcp, messaging, notes, notion, office, rust, slack, slack-alternative, startup, startup-tools, typescript, workspace
A Rust-based unified workspace combining email, chat, docs, tasks, and AI agents with shared memory, targeting team productivity.
megadose/holehe
⭐ 12,498 stars · +195 today · Python · topics: ebay, email, emails, information-gathering, instagram, open-source-intelligence, osint, osint-python, osint-tools, pypi, python, social-network, tellonym, trio, twitter
Command-line tool to check if an email address is registered across multiple social platforms using forgotten password flows, aiding OSINT.
smicallef/spiderfoot
⭐ 20,712 stars · +283 today · Python · topics: attacksurface, cti, cybersecurity, footprinting, hacking, information-gathering, information-security, infosec, intelligence-gathering, osint, osint-framework, osint-reconnaissance, osint-tool, pentesting, python, recon, security-tools, threat-intelligence, threatintel
Automates open source intelligence gathering for mapping attack surfaces and cybersecurity threat intelligence.
NVIDIA-NeMo/Switchyard
⭐ 1,291 stars · +408 today · Rust
Enables flexible traffic routing across multiple LLMs and providers while preserving compatibility with OpenAI and Anthropic APIs for benchmarking and cost optimization.
Source: GitHub Trending — Daily [1]