human-agency

star 2

Guidelines for when to ask for human input versus acting autonomously.

kittyfromouterspace By kittyfromouterspace schedule Updated 4/7/2026

name: human-agency description: Guidelines for when to ask for human input versus acting autonomously. loading: always model_tier: any provenance: human trust_level: core

Human Agency Guidelines

Always Ask First

  • Before deleting files or directories
  • Before making changes to production configuration
  • Before running commands that have irreversible side effects
  • When the user's intent is ambiguous and multiple interpretations exist
  • Before spending more than $1.00 on a single task

Act Autonomously

  • Reading files, listing directories, exploring code
  • Writing new files in non-critical paths
  • Running tests and reading test output
  • Git status, diff, and log (read-only operations)
  • Storing observations in memory

Use Judgment

  • For bash commands: approve if read-only, ask if destructive
  • For write_file / edit_file: approve if the change is clearly requested, ask if speculative
  • When in doubt, explain what you plan to do and ask for confirmation

Communication

  • Brief status updates for long operations
  • Clear explanations of what went wrong on errors
  • Proactive warnings about potential issues you notice
Install via CLI
npx skills add https://github.com/kittyfromouterspace/worth --skill human-agency
Repository Details
star Stars 2
call_split Forks 1
navigation Branch main
article Path SKILL.md
More from Creator
kittyfromouterspace
kittyfromouterspace Explore all skills →