Skip to content
Get packs

Agent-Operated Software

Guides an agent in designing, building, and repairing apps whose live backend includes OpenRig agents

Source Download skill
Repository
mvschwarz/openrig
Stars
8,553
Licence
Apache-2.0
Files
1 file

What it does

This skill gives an AI agent a playbook for building and maintaining an ongoing application where AI agents are part of the running backend, not just a helper. It explains how to keep behavior in Markdown, YAML, or JSON control files, how to make generated artifacts verifiable (write to a partial path, verify, rename, then write a completion marker), and where to fix problems (in the source every agent reads, not in a copied file). It is written as guidance for the agent rather than a tool that performs actions by itself.

Great for

  • Developers building an agent-backed app on top of OpenRig who need a shared architecture vocabulary
  • Teams diagnosing why an agent-driven application behaves unexpectedly and need to find the layer at fault
  • People designing a thin UI over specialist agent roles with Markdown or YAML state

Skip it if

  • Simple one-off automations or single bounded tasks; the skill itself distinguishes these as Agent-Operated Workflows, which need a different skill (agent-operated-workflows)
  • Projects not using OpenRig; the guidance assumes OpenRig agents, rigs, and the rig CLI, so it adds little elsewhere
  • Core OpenRig substrate bugs; the skill explicitly says those require full reproduction and code fixes, not the looser coherence-first approach

Try asking

  • “Help me design an app where a Markdown control plane drives a team of specialist agents”
  • “Diagnose why my agent-backed dashboard shows stale state; trace whether the bug is code or a skill gap”
  • “Review my build pipeline so generated video files are only consumed after a .done sentinel is written”
  • “Explain how to split scope, item judgment, and workflow publishing in my agent-operated project”
Step-by-step guide: how to use Agent-Operated SoftwareInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 1 original files, the Apache-2.0 licence and an attribution note.

    agent-operated-software.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip agent-operated-software.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip agent-operated-software.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Install the sibling skills it references, since its guidance routes to them for exact procedures
  • Point the agent at a concrete project folder and ask it to trace the path that produced a behavior before editing anything
  • Run rig proof --help as the skill suggests to learn the supported judgment and read commands

What's inside

  • SKILL.md

Redistributed unmodified from mvschwarz/openrig under the Apache-2.0 licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.