Last Updated: 1 June 2026
Public-safe reusable toolkit for organising AI-agent rules, n8n workflow helpers, CI/CD setup patterns, copyable skills, and source-preserved project modules into one validated source-of-truth repo.
<aside> 🔗
https://github.com/weijunswj/ai-agent-toolkit
</aside>
AI Agent Toolkit is a reusable source-of-truth repo for AI-assisted development assets.
It came from a very practical problem: I started having too many separate repos for different things.
One repo had n8n setup notes. Another had CI/CD prompts. Another had workflow helpers. Others had agent rules and experiments. It worked at first, but once I needed to reuse or reference something, finding the right source became annoying.
That led me to build a central Skills hub: one place where AI-agent rules, reusable guides, workflow helper patterns, and project assets could live in a cleaner structure.
The project became a simpler and more reliable source-of-truth system — a public-safe toolkit repo that preserves source material, publishes copyable skills, and validates that generated outputs do not drift from the source.
After the repo-wide cleanup, the toolkit is intentionally skills-first: humans work from _projects/**, AI agents consume skills/**, and generated outputs are checked against their declared source contracts.
AI-agent tooling gets messy fast once it spreads across too many repos and notes.
The main issues were: