Daily Digest – 2026-03-29

Daily curated repos and tools for solo founders, AI/B2B SaaS builders, and small business tools. 187 repos evaluated, 16 made the cut. Plus 36 picks from HuggingFace, Replicate, Papers with Code, npm, PyPI, GitLab, Ollama, and launch platforms.

SaaS starters and templates

  • Skolaczk/next-starter – Next.js starter with everything (Stripe, auth, tests). Skip boilerplate, build faster. (TypeScript, 974★)

  • agustinusnathaniel/nextarter-chakra – Next.js + Chakra UI + TypeScript starter. Get a sleek UI project going quickly. (TypeScript, 839★)

  • windingwind/zotero-plugin-template – Zotero plugin template. If you use Zotero, build your own integrations easily. (TypeScript, 801★)

  • daltonmenezes/electron-app – Electron app boilerplate with React 19, Tailwind 4. Get desktop apps shipped, fast. (TypeScript, 796★)

  • scaffdog/scaffdog – Markdown-driven scaffolding tool. Automate repetitive file generation like a pro. (TypeScript, 763★)

AI agents, LLM infra and RAG

UGC, social media and creator tools

  • AiuniAI/Unique3D – Generate high-quality 3D meshes from a single image. Create assets without a 3D artist. (Python, 3.6k★)

  • open-xml-templating/docxtemplater – Generate docx, pptx, xlsx from templates. Automate report generation, save hours. (JavaScript, 3.5k★)

  • Tencent-Hunyuan/Hunyuan3D-1 – Unified framework for text-to-3D and image-to-3D. Create 3D assets from text or images. (Python, 3.5k★)

  • DLR-RM/BlenderProc – Procedural Blender pipeline for photorealistic images. Generate training data efficiently. (Python, 3.5k★)

  • thu-ml/TurboDiffusion – 100–200× acceleration for Video Diffusion Models. Make video AI models actually usable. (Python, 3.4k★)

  • binhnguyennus/awesome-scalability – Patterns for scalable, reliable, performant systems. Learn how to not scale badly later. (Multi, 70k★)

  • Z4nzu/hackingtool – ALL IN ONE Hacking Tool For Hackers. Probably don’t need this, unless you’re that kind of builder. (Python, 55.9k★)

  • jellyfin/jellyfin – Free Software Media System. Self-host your media, ditch commercial services. (C#, 49.7k★)
  • upscayl/upscayl – Free and Open Source AI Image Upscaler. Improve low-res images for free, finally. (TypeScript, 44.3k★)

Replicate Models

Ollama Models

  • ministral-3 – Ministral 3 for edge deployment. Run powerful LLMs locally on cheaper hardware. (772.8K pulls)

  • codeqwen – CodeQwen1.5, LLM trained on code. Get better code generation and analysis from an LLM. (770.2K pulls)

  • openhermes – OpenHermes 2.5, 7B model. A strong, fully open-source LLM for various tasks. (769.7K pulls)

  • qwen2-math – Qwen2 Math specialized LLM. If your AI needs to do math well, this is for it. (756K pulls)

  • phi3.5 – Lightweight 3.8B AI model. Get powerful performance from a small, efficient model. (755.1K pulls)

npm Packages

  • scrivito – SaaS Enterprise CMS for agencies. If you need a managed CMS for big clients. (22K/wk)

  • @m2d/core – Markdown to DOCX core engine. Convert Markdown to Word documents effortlessly. (21K/wk)

  • @genkit-ai/compat-oai – Genkit AI plugin for OpenAI APIs. Integrate OpenAI models into Genkit projects. (19K/wk)

  • @m2d/html – Extend MDAST by parsing HTML in Markdown. Convert complex Markdown with HTML to DOCX. (18K/wk)

  • mdast2docx – Convert Markdown AST to DOCX. Generate Word docs from structured Markdown content. (18K/wk)

PyPI Packages

  • dspy-ai – DSPy library. Program LLMs instead of just prompting them, smarter AI pipelines. (488K recent downloads)

  • Crawl4AI – LLM Friendly Web Crawler & scraper. Get clean data for your AI models, easily. (376K recent downloads)

  • outlines – Probabilistic Generative Model Programming. Control LLM output format precisely, critical for agents. (358K recent downloads)

  • pinecone-client – Pinecone client (DEPRECATED). Reminder: This vector database client is deprecated, migrate. (336K recent downloads)

  • replicate – Python client for Replicate. Run powerful AI models with a few lines of Python. (297K recent downloads)

HuggingFace Spaces

  • tonyassi/voice-clone – Voice cloning space. Try out voice synthesis, quickly. (gradio, 2613 likes)

  • fffiloni/diffusers-image-outpaint – Image outpainting space. Extend your images, experiment with AI creativity. (gradio, 2513 likes)

  • suno/bark – Bark text-to-audio model space. Generate realistic speech and audio, effortlessly. (gradio, 2370 likes)

  • sczhou/CodeFormer – CodeFormer face restoration space. Fix blurry faces in your images. (gradio, 2351 likes)

  • microsoft/HuggingGPT – HuggingGPT space. Orchestrate multiple AI models for complex tasks. (gradio, 2270 likes)

HuggingFace Models

HuggingFace Datasets

New Launches