User guide

A guided tour of Rune, in reading order. This page is the connective tissue: it explains the shape of the tool and points you to the how-to guide or reference for each capability, rather than repeating them. Read it top to bottom once; come back to the links as you need them.

1. Start here

If you’re brand new, do these two first — they’re the fastest path to a working task:

1. What is Rune? — the idea, the mental model, and when to reach for it.
2. Getting started — install, write a tiny Runefile, run a task.

2. The core model

A Runefile is named blocks of commands you run by name, with declared dependencies between them. Everything else builds on that. For the complete syntax in one place, keep the Runefile language reference open.

3. Composing work

• Dependencies & post-hooks — order work; run things before, after, and on failure.
• Parameters — defaulted, required, and variadic inputs.
• Parallelism — run independent prerequisites concurrently.
• Imports & modules — split and namespace task files across a repo.

4. Controlling how tasks run

• Executors — shell, Python, Node, or an AI agent as the task body.
• Caching — opt-in, content-hash skipping for expensive tasks.
• OS filtering — scope a task to an operating system.
• Settings & dotenv — project settings and .env files.

5. Sharing tasks with AI agents

Rune is AI-native: tasks are first-class MCP tools, read-only by default and gated when destructive. See the AI agents (MCP) reference and the MCP walkthrough.

6. Putting it together

When you’re wiring up a real project, jump to the use-case walkthroughs (Python, Node, AI agents) — each shows a complete Runefile built from the pieces above.

[!TIP] Hit an error? Rune’s diagnostics point at the exact file:line:col. The Troubleshooting guide maps common messages to fixes.

---

Next: How-to guides · Use cases · CLI reference · Runefile language