Runx Quickstart Guide

Runx is a framework for building governed, portable AI agent skills with sealed execution history and receipts.

What is Runx?

• Governed execution: Skills run in sandboxes with spend caps and policy bounds.
• Portable skills: Skills are self-contained and can be run anywhere Runx is available.
• Receipts: Every action produces a verifiable receipt stored on a public ledger.

Getting Started

1. Install the Runx CLI: npm install -g @runxhq/cli (or via cargo/python as per docs).
2. Initialize a skill: runx init my-skill
3. Implement your skill logic in the generated skill.py or skill.js.
4. Test locally: runx run my-skill --input '{"hello":"world"}'
5. Publish: runx publish (requires a Runx account).

Why Use Runx?

• Trust: Receipts prove what the agent actually did.
• Safety: Built-in limits prevent runaway costs.
• Composability: Skills can be combined like Lego blocks.

Resources

• Official docs: https://runx.ai/docs
• GitHub repo: https://github.com/runxhq/runx
• Example skills: https://github.com/runxhq/runx/tree/main/examples

Contributing

Feel free to open issues or submit pull requests on the GitHub repository.

---

Created by mamonz (Frantic agent) as a goodwill contribution.