karimaballanews

The weekly · Week 35 · 24 Aug – 30 Aug 2026

AI coding agents: what changed

This week's items center on coding agents: terminal agents, skill collections, plugin directories and prompt libraries.

8 picks

01GitHub★ 127,935

OpenAI Codex: A Terminal Coding Agent

A lightweight coding agent that runs in your terminal. It is aimed at developers who want an AI coding assistant on the command line.

Image: GitHub

02GitHub★ 3,834

Unlazy Skill Targets Unfinished Agent Work

A skill for AI agents that splits a task into layers, the Depth Tree method, so each part gets the full time budget. It aims to stop unfinished work.

Image: GitHub

03GitHub★ 35,236

Curated List of 1000+ Agent Skills

A curated collection of 1000+ agent skills from official dev teams and the community. It works with Claude Code, Codex, Gemini CLI, Cursor and more.

Image: GitHub

04GitHub★ 918

Cybermes: Offensive Security Agent Framework

An agent framework for offensive security, bug bounty and red teaming, built on Hermes Agent with multi-model LLM orchestration. It is aimed at security researchers.

Image: GitHub

05GitHub★ 37,427

Official Directory of Claude Code Plugins

An official directory, managed by Anthropic, of high-quality Claude Code plugins. It helps developers find plugins for their coding agent.

Image: GitHub

06GitHub★ 4,475

Community Plugin Marketplace for Claude Code

A community plugin marketplace for Claude Code and Claude Cowork. It is a read-only mirror; plugins are submitted through the official plugin directory submission page.

Image: GitHub

07GitHub★ 33,965

GPT Image 2 Prompt and Case Library

A library of GPT Image 2 prompts with 530+ cases, 20+ templates and reusable skills. It helps people write image prompts with more predictable results.

Image: GitHub

08DEV

AWS DevOps Agent Built with Kiro Crew

A write-up on connecting Kiro Crew to an AWS DevOps agent over MCP for incident detection, investigation and fixes. One config block exposes 34 tools.

Picked from public sources (GitHub, Hacker News, DEV, Lobsters, Reddit, Product Hunt) by our daily scout. Each summary uses only what the source itself says; GitHub stars are as checked on the date shown.