keli-wen

keli-wen / agentic-harness-patterns-skill

Public

Agent skill for harness engineering — memory, permissions, context engineering, multi-agent coordination. Distilled from Claude Code, with Codex CLI and Gemini CLI on the roadmap. EN/ZH. Install via npx skills add.

agent claude-code codex contexts gemini-cli
34
14
100% credibility
Found Apr 02, 2026 at 34 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
AI Summary

This repository shares practical design patterns for building dependable AI agent systems, drawn from studying Claude Code's approach and offered as easy-to-read guides and add-ons in English and Chinese.

How It Works

1
🔍 Discover the Guide

You find this collection of smart tips for making AI helpers work reliably while browsing for ways to build better AI assistants.

2
📖 Explore the Lessons

You read the welcoming page that explains six key areas like memory, safety, and teamwork to solve real-world AI challenges.

3
Pick Your Way
👀
Study Alone

Flip through the detailed stories and tips in English or Chinese to learn at your leisure.

➕
Add to Helper

Connect the ready-made lessons to your AI tool so it can draw on these patterns whenever needed.

4
💡 Unlock Key Insights

You grasp powerful ideas for keeping AI focused, safe, and coordinated, feeling empowered to tackle tough problems.

5
🛠️ Put Ideas to Work

Apply the patterns to your own AI projects, watching them become steadier and smarter with each tweak.

🎉 AI Agents Thrive

Your AI helpers now handle big tasks smoothly, stay secure, and grow without headaches, just like the pros.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 34 to 34 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is agentic-harness-patterns-skill?

This GitHub agent repo delivers portable design patterns for building reliable AI agent harnesses—the infrastructure wrapping LLM-tool loops with memory, permissions, context control, and multi-agent coordination. Distilled from Claude Code's production runtime, it solves why simple agent loops fail at scale: forgotten sessions, exploding contexts, unsafe tools, and chaotic swarms. Install via `npx skills add` into the agent skills ecosystem, pulling English or Chinese versions for Claude Code, upcoming Codex CLI, or Gemini CLI.

Why is it gaining traction?

Unlike vague agent skills examples or prompt tutorials, it grounds patterns in Claude Code evidence without code dumps, making them runtime-agnostic for any agent github copilot vscode or cli setup. Developers hook into the agent skills marketplace for lazy-loaded skills that front-load triggers and fail-closed safety, skipping re-explaining workflows. Bilingual docs and a clear "problem -> rules -> gotchas" structure make it instantly actionable over scattered agent github copilot reddit threads.

Who should use this?

Backend engineers crafting custom agent github actions or multi-agent orchestration for production coding agents like Cursor or Windsurf. Teams extending agent github copilot intellij plugins needing context budgets that don't explode. Anyone building agent skills io libraries or github agent repos where LLMs call tools reliably across sessions.

Verdict

Grab it for educational depth on harness patterns—docs are polished, MIT-licensed, and roadmap promises cross-runtime synthesis—but with 34 stars and 1.0% credibility score, treat as early inspiration, not battle-tested code. Solid for agent skills library builders prototyping safely.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.