feed · watching · stats · help with your tokens
AI & LLMAgents & MCPSelf-hostedNetworkingSecurityDevOps & cloudData & databasesWebMobileTerminalDeveloper toolsLearningMusic & mediaHome & IoT
PythonTypeScriptJavaScriptGoRustCC++JavaC#SwiftKotlinPHPRubyShellHTMLZig
finic-ai/kairos · 14h
Kairos is an open-source, real-time fraud and risk decisioning platform for sign-ups, logins, orders, payments and payouts.
It is an open-source alternative to commercial fraud tools, adding an agent that tunes it and decision models that work before labels exist. It starts its own Postgres, so there is no separate database to set up.
Hacker News · Show HN: Kairos – Open-source risk and fraud decisioning platform · 95 points
Muvon/octomind · 20h
Octomind is an open-source coding agent harness that gives an LLM real access to a repository through MCP tools.
It is built for agent work that keeps going after the human leaves the screen, enforcing policy as code instead of approval popups. One Rust engine with one TOML config serves an interactive CLI, stdin, a background daemon, WebSocket and an ACP agent.
Hacker News · Octomind – open-source AI harness that saves your tokens, static rust · 54 points
NVIDIA/auto-ontology · 28h
An open context layer that turns database metadata into an ontology, so AI agents answer questions about enterprise data with fewer wrong answers.
It gives agents one trusted definition per business concept, mapped across tables and databases, so analytics answers are grounded instead of improvised. Agents query through the ontology, and people review and certify it in a web UI; terms, reviewed SQL and joins are served over MCP and a REST API.
Hacker News · Nvidia – open-source auto-ontology · 281 points
KKKKhazix/agent-autoloop · 27h
Agent Autoloop lets a command-line agent pick up work in the background and follow it through, asking for a decision only when something is uncertain.
It addresses command-line agents that only work while someone watches the terminal, so the job stops once the window is closed or the person walks away. A kernel of about 1650 lines of code with zero runtime dependencies coordinates the work.
GitHub · New on GitHub · 1,893 stars
aaitor/skill-video-studio · 41h
A local video-production workspace driven by a Claude Code skill that turns a product walkthrough into a polished demo video.
It turns real product interactions into demo videos, from browser capture and title cards to a mastered and inspected file. The skill coordinates existing tools rather than replacing them: Playwright for capture, HyperFrames for composition and FFmpeg for mastering.
Hacker News · Skill to Create Product Videos · 118 points
claude-mem preserves context across coding-agent sessions by capturing tool usage and generating semantic summaries.
It stops an agent from forgetting a project between sessions by capturing tool observations, summarizing them and recalling them later. It runs as a local worker service with a web viewer and search endpoints, managed by Bun.
D4Vinci/Scrapling · 8h
Scrapling is a Python web scraping framework that combines fast HTTP requests, stealth browsers, parsing and concurrent crawling in one library.
It covers the full scraping workflow, from extracting a few fields to collecting a whole website as clean Markdown, from Python, a terminal or an AI agent. It combines fast HTTP requests, stealth browsers, JavaScript rendering and Cloudflare challenge handling behind one interface.
JustVugg/colibri · 8h
colibri is a C inference engine that runs very large open mixture-of-experts models on the machine you already have.
It makes models of this size usable on hardware people already own, and keeps the engine small enough that anyone can measure and speed it up. It is written in pure C, with one file per model family, and needs no GPU.
GitHub · A hierarchical inference engine for frontier MoE models
vercel-labs/agent-desktop · 8h
agent-desktop gives AI agents computer use on macOS, Windows and Linux by reading the OS accessibility tree instead of screenshots.
It lets an agent operate desktop applications through element references rather than pixel matching, while the control loop stays in the agent. A CLI and a C library share one core that reaches the operating system only through a platform adapter.
Balanov/uncut · 2d
The lossless, headless coding agent harness. One 7 MB static binary, four tools, plain stdin/stdout, no silent cuts.
GitHub · New on GitHub · 599 stars
deepseek-ai/dsh-superagent-prototype · 4d
dsh-superagent is a research prototype for a proactive coding agent that works on a single GitHub repository using DeepSeek Harness.
It aims to maximize value while minimizing human attention by acting only when asked or permitted by team rules, serving developers who want autonomous coding assistance without constant oversight. It listens to GitHub and Feishu, opens work items, executes tasks in isolated agent sessions, and responds with pull requests or conversation updates.
GitHub · New on GitHub · 18,092 stars
zhanghao123vip/ai-manju-skills · 4d
A collection of 21 genre-specific templates for generating full AI comic drama production workflows, including scripts, prompts, and release assets.
It solves the problem of structuring AI-generated comic drama production by providing a complete workflow from a single story idea to final release assets for creators using AI tools. The system uses standalone SKILL.md files that require no dependencies and can be used directly in Claude Code, WorkBuddy, or pasted into chat tools like DeepSeek.
GitHub · New on GitHub · 568 stars
wangzan101/Mika · 5d
An on-device AI agent for Android that operates apps and the web to get things done for you.
GitHub · New on GitHub · 6,364 stars
FlashProve is a research prototype for zero-knowledge proofs of AI inference that allows providers to prove model identity without revealing weights.
It solves the problem of verifying AI model identity for providers by allowing them to prove which model generated a response without exposing the underlying weights, ensuring trust without compromising intellectual property. The protocol's soundness is proved in Lean 4, ensuring that a provider passing off non-committed output is caught with high probability.
GitHub · New on GitHub · 655 stars
N1nEmAn/sapper-v2 · 4d
Autonomous pentest legion — AI commander + worker swarm + layered fact-intent graph. 3rd Tencent Intelligent Attack-Defense online champion (634 teams · 8h unattended · 1B tokens beat 4.5B). The Nüwa Test: give AIs Create/Language/History, let a civilization grow.
GitHub · New on GitHub · 687 stars
sleepy-zone/ai-ref-library-cn · 5d
AI 作图素材库 · 中文版（脸 / 姿势 / 服装）
GitHub · New on GitHub · 712 stars
wanghao-SunSky/cs329z-chinese · 6d
斯坦福 Stanford CS329Z Engineering AI Agents 中文教程：面向 Python 开发者的 AI 智能体课程，19 讲与 4 项实践，涵盖 LLM、RAG、工具调用、记忆、多智能体、评测与安全。完整 Markdown、在线与离线 HTML。Chinese course notes.
GitHub · New on GitHub · 492 stars
That is everything from the last 30 days.