📦 win4r/clawteam-openclaw

ClawTeam OpenClaw: The Multi-Agent Swarm Coordinator

ClawTeam OpenClaw is the multi-agent swarm coordinator that lets your AI coding agents build a team and divide the work. It takes a single goal and automatically spawns multiple workers, giving each one their own isolated space to code without stepping on the others' toes. The agents handle the heavy lifting by splitting tasks, sending messages to each other, and merging the finished results, so you never have to micromanage the process. It works with OpenClaw by default and supports other tools like Claude Code, making it easy to turn a single bot into a collaborative team.

🆔 @hackernewsgithubprojects

📦 shenhao-stu/openclaw-agents

OpenClaw Agents: One Command for an AI Research Team

Provision a full team of specialized A.I. agents to handle complex research tasks in your chat. OpenClaw Agents automates the setup of nine distinct A.I. personalities that collaborate to write and review academic papers. The standout feature is the adversarial collaboration built into the system. A creative ideator generates a concept, but a strict critic agent acts as a quality gate, rejecting weak ideas until the proposal meets a high standard. This creates productive tension that improves the final output. You can deploy this entire fleet to messaging apps like Telegram or Discord with a single command. It is a fascinating look at how structured A.I.

🆔 @hackernewsgithubprojects

📦 meterless/meterless

Meterless: Your Computer Works While You Sleep

Your Mac or DGX Spark becomes an AI hotspot that works while you sleep. Meterless runs local models overnight to attempt hundreds of versions of a task, checking each one against strict criteria. It races through these attempts in the dark, discarding failures and keeping only the results that pass every test. You wake up to finished files and a receipt verifying exactly which checks passed. This solves the problem of waiting for AI to finish big jobs, turning idle hardware time into reliable, verified output without cloud costs.

🆔 @hackernewsgithubprojects

📦 tangency-labs/fieldcraft

FieldCraft: The AI-Driven Engineering Tool

An AI agent can now open a 3D part, generate a complex mesh, run a full structural analysis, and verify the results against physics, all from a single prompt. That capability lives in FieldCraft, an open-source engineering application built entirely in pure Rust. Instead of relying on expensive, closed-source software, this project gives engineers a free tool that handles everything from meshing to solving for heat and fluid flow. The standout feature is its built-in verification system, which automatically checks every calculation against known theoretical answers to ensure accuracy.

🆔 @hackernewsgithubprojects

📦 perrylink/dsh-permission-rules

How to Control AI Tools

You can write a simple text file to stop an artificial intelligence from deleting your hard drive before it even tries. DSH Permission Rules acts like a strict bouncer for AI coding assistants, checking every tool call against a list you control. If a command looks dangerous, the system blocks it instantly, asks you for permission, or lets it pass without hesitation. This is the same permission style used by Claude Code, bringing predictable safety to the DeepSeek Harness. You define exactly which files and commands are safe, and the system enforces that policy in real time.

🆔 @hackernewsgithubprojects

📦 opencoredev/akeru-bot

Akeru Bot: Your Own AI Team

Akeru Bot lets you run named AI teammates on your own machine, giving you full control over who does what. It is a self-hosted tool that works on desktop, web, and mobile, so you are not stuck relying on someone else's servers. The project keeps your conversations and settings in a private folder on your computer, meaning no data leaks to outside companies. It connects to major coding models, but you decide exactly how each bot behaves and what it remembers. This is great for developers who want personalized help without giving up their data.

🆔 @hackernewsgithubprojects

📦 stepfun-ai/step-5-preview

Step 5 Preview: The AI That Actually Works

Step 5 Preview is an AI model that doesn't just chat; it actually does the work. It's designed to handle complex, multi-step tasks like writing code or analyzing financial reports, acting like a tireless assistant that remembers everything you've discussed. Unlike simple chatbots that forget context after a few pages, this model can process a massive amount of information in one go, keeping track of the big picture as it solves problems. It’s built to be a genuine partner for developers and analysts, handling the tedious, repetitive parts of a job so you can focus on the creative stuff.

🆔 @hackernewsgithubprojects

📦 orchestratorinc/agent-orchestrator

Agent Orchestrator: The Control Tower for Coding AI

A single dashboard can supervise over thirty different coding AI agents working on your project at once. The tool called Agent Orchestrator turns the chaotic process of running multiple AI workers into a manageable team. It gives every task its own isolated workspace and branch so agents do not collide, while a central planner breaks big goals into smaller jobs. You watch every pull request, test result, and review update on one live board. It is no longer about prompting one chatbot, but about managing a whole crew of digital employees.

🆔 @hackernewsgithubprojects