How to use the OpenRig Skills Index agent skill

What is an agent skill, and what does this one do?

An agent skill is a small document that tells an AI agent when to do something and how to do it. The agent sees each skill's name and description all the time, and opens the full body only when the description matches what it is doing.

The openrig-skills skill is the map of OpenRig's own skills. It does not perform engineering tasks itself. It points the agent to the documents that do, covering moments such as fleet recovery, seat handover, new-seat orientation, a watchdog wake, cross-host reach to an agent on another machine, systematic debugging, queue triage, and implementation planning.

• Helps an agent that does not know which procedure applies.
• Helps an agent when nothing is projected into its context on a cold boot.
• Loads the chosen skill with the rig command-line tool rather than by guessing file paths.

Check the requirements before you install

The skill only makes sense inside an OpenRig environment. The source requires that OpenRig is installed and that the rig command-line tool is available to the agent. The skill's allowed tools are limited to Bash(rig:*), so the agent can run rig commands only.

• OpenRig installed, with the rig command-line tool available to the agent.
• Bash access restricted to rig commands (allowed-tools: Bash(rig:*)).
• Not for people outside OpenRig or who do not run agent teams.
• Installing or upgrading OpenRig itself belongs to a separate rigs skill, not this one.

Install the openrig-skills skill

Download the ZIP for the openrig-skills skill from the directory page, then unzip it. The skill's files include SKILL.md. Place the unzipped skill folder in one of these locations:

You can also install it through the Claude apps: open Settings, then Capabilities, then Skills, and upload the skill there.

• Personal install: ~/.claude/skills/ (unzip so the openrig-skills folder sits inside it).
• Project install: a project's .claude/skills/ folder.
• Claude apps: Settings → Capabilities → Skills, then upload the skill.

How to trigger the skill with natural requests

You do not need a special command. The agent matches your request to the skill's description. The skill is aimed at moments inside an OpenRig team, so phrase requests around those moments.

When the agent does not know which procedure applies, it should open this index first. The index tells it to discover the shipped entries with rig context list, pick the one whose when matches, and load it with rig context get.

• "I don't know which OpenRig skill applies to this. Start with the index."
• "Nothing was projected into my context after the cold boot. What should I load?"
• "I'm taking over this seat from another agent. Which skill covers the handover?"
• "A seat on another machine isn't responding. Which skill handles cross-host reach?"
• "Queue work looks stuck. Which procedure should I open?"

How loading works: discover, select, load

The skill uses three steps. The agent runs rig context list to see each entry's reference and name, then matches its moment to the right row and takes that row's reference. Finally it loads the entry with rig context get. A unique bare name such as watchdog also resolves, but a reference containing a slash is an exact lookup and fails if it does not exist.

The content is served from the installed CLI, so what the agent loads matches the running version of OpenRig.

• Discover: rig context list (or rig context list --json for each entry's purpose).
• Select: match the moment to a skill's when and take its reference.
• Load: rig context get with the full reference.

Examples of what the index points to

Each row in the index names a skill and the moment it applies to. The examples below come from the index itself. Note that some skills ship in the repo and only reach a seat when its profile selects them.

• rig-lifecycle: rig up, down, pause and resume operations.
• watchdog: monitoring or recovering a seat with health, restore or stuck-state work.
• cross-host-rig-commands: reaching seats or queues on another host using --host.
• openrig-upgrade: upgrading OpenRig or the daemon.
• rig-bundles-and-shareable-artifacts: packaging or installing a rig bundle.
• systematic-debugging: loaded with rig context get skills/process/systematic-debugging.

Safety notes and tips

The source states that this skill runs no scripts and uses no network. Its only tool permission is rig commands. Content is served from the installed CLI rather than fetched from the internet.

Some skills it points to can change live state, such as rig lifecycle, topology and seat management, so check that the agent is acting on the right rig before following a procedure.

If you use the skill, keep the description accurate for your agents so the matching stays reliable.

• Runs scripts: no. Uses network: no.
• Allowed tools are limited to Bash(rig:*).
• Use rig context list --json when you want each entry's purpose in structured form.
• If a slash-bearing reference fails, it does not exist. Check the list rather than guessing paths.