Skip to content
RuneRune

Executors

How to write task bodies in shell, Python, Node, or an AI agent. Part of the guides; full syntax in the language guide.

Concept

Section titled “Concept”

A task body runs under an executor named in parentheses after the signature. The default is sh — a pure-Go, cross-platform shell (mvdan.cc/sh) that behaves the same on Linux, macOS, and Windows without invoking the system shell. Other executors (python, node, custom) shell out to the real interpreter via a temp file; agent drives an installed AI-agent CLI.

Syntax

Section titled “Syntax”
# default (sh) — no parentheses needed
hello:
    echo "hi"


# Python (needs python3 installed)
analyze (python):
    print("analyzing")


# Node (needs node installed)
bundle (node):
    console.log("bundling")


# Agent — a natural-language instruction (see the MCP guide)
summarize (agent):
    Summarize the latest git changes.

Runnable example

Section titled “Runnable example”

See examples/polyglot (shell + Python + Node in one pipeline), plus node-project and python-project.

Pitfalls

Section titled “Pitfalls”
  • Interpreters must be installed. A (python)/(node) task fails with an actionable error if the runtime isn’t on PATH. The Rune container image ships only the static binary — see the Docker guide for that caveat.
  • Stick to the default sh for portability. Only reach for set shell := [...] when you truly need a system shell feature; it trades away cross-platform parity. See shell override for how an override runs the body.
  • Agent bodies need a configured agent CLI — see AI agents (MCP).

Next steps

Section titled “Next steps”