Agent-Skills.md

Agent Skills: Moonrepo Run Tasks

Runs moonrepo tasks with project/target selectors and troubleshooting flow, when daily execution needs predictable scoped runs, resulting in faster feedback and fewer ad hoc command mistakes.

UncategorizedID: zenobi-us/dotfiles/run-tasks

Repository

zenobi-us/dotfiles
zenobi-usLicense: MIT
676

Install this agent skill to your local

pnpm dlx add-skill https://github.com/zenobi-us/dotfiles/tree/HEAD/devtools/files/pi/agent/bundles/platform/skills/devtools/moonrepo/run-tasks

Skill Files

Browse the full folder contents for run-tasks.

Download Skill

Loading file tree…

devtools/files/pi/agent/bundles/platform/skills/devtools/moonrepo/run-tasks/SKILL.md

Skill Metadata

Name
run-tasks
Description
Runs moonrepo tasks with project/target selectors and troubleshooting flow, when daily execution needs predictable scoped runs, resulting in faster feedback and fewer ad hoc command mistakes.

Moonrepo Run Tasks

Overview

Use this skill for routine execution of moonrepo tasks by project and target, with selector-driven filtering and failure triage.

When to Use

  • Running one project target quickly.
  • Running affected subsets instead of whole repo.
  • Diagnosing common moon task execution failures.
  • Running tasks in CI

Core Command Patterns

  1. Single project target 
    moon run app:build
moon run api:test
  2. Multiple targets/projects 
    moon run :lint
moon run app:build api:test
  3. Affected/scope workflows 
    moon run :test --affected
moon run :lint --projects app,api
  4. Inspect before running 
    moon query projects
moon query tasks app

Running Commands in CI

when running tasks in CI, use ci instead of run.

moon ci :lint
moon ci app:build api:test

moon query projects
moon query tasks app

Triage Flow

  1. Unknown target/project
    • Run moon query projects and moon query tasks <project>.
  2. Unexpected dependencies or task fan-out
    • Re-run with explicit project/target pair.
    • Inspect task config and inheritance in .moon/ + project configs.
  3. Tool/runtime mismatch
    • Confirm runtime versions via project toolchain manager before retry.
  4. Cache confusion
    • Capture run output first, then compare with clean run (--updateCache only when needed by your policy).

Caveats

  • Prefer selectors over broad moon run :target when repo is large.
  • Avoid “just run everything” in CI-like local checks; it hides dependency problems.
  • Always verify project and target names from query commands; don’t guess.

Quick Examples

# validate only one changed app
moon run app:test --affected

# lint two projects only
moon run :lint --projects app,dashboard