Guide · Agent skill
How to use the Rig Lifecycle Operations agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The rig-lifecycle skill is a set of written instructions, a SKILL.md file, that teaches an AI agent how to reason about OpenRig rigs: groups of AI coding agents run as managed topologies. It covers operations such as create, start, stop, resume, restore, snapshot, release, unclaim and destroy. This guide explains what the skill changes about the agent's behaviour, how to install it, how to trigger it, and what to watch for. It assumes you have never used an agent skill before.
Rig Lifecycle Operationsmvschwarz/openrig · Apache-2.0 Download the skill Quick start
- 1Install OpenRig. Confirm the rig CLI and daemon are installed, since the skill refers to them and cannot operate without them.
- 2Download and unzip the skill. Download the ZIP from this page and unzip it so that SKILL.md sits inside a rig-lifecycle folder.
- 3Place the skill folder. Move it into ~/.claude/skills/ for personal use or into .claude/skills/ in a project, or upload it under Settings → Capabilities → Skills in the Claude apps.
- 4Describe your lifecycle task. Ask in plain language about a start, resume, restore, reboot, or rig ps question, and name the rig if you have one.
- 5Check the reported outcome. Look for resumed, rebuilt, fresh, failed or attention_required, and confirm failures are reported loudly rather than replaced by a fresh start.
- 6Verify against the runtime. Do not accept rig ps output alone after recovery. Ask the agent to confirm the actual runtime state.
What an agent skill is and what this one changes
An agent skill is a folder containing a SKILL.md file with instructions that an AI agent loads when a task matches the skill's description. It adds guidance; it does not add a program that runs on its own.
The rig-lifecycle skill makes the agent reason carefully about OpenRig lifecycle operations. It tells the agent to treat output from rig ps and lifecycle projections as summaries rather than ground truth after a reboot or recovery, and to check the runtime itself. It also tells the agent that a failed resume must be reported as FAILED loudly, with no automatic fresh-launch fallback. Fresh launch is an explicit follow-up step that a person chooses.
- Knows the lifecycle operation family: create, start, stop, resume, restore, snapshot, release, unclaim, destroy.
- Warns about four failure modes: partial auto-restore, projections that look healthier than reality, provider auth problems treated as code work, and resume working for one runtime but failing for another.
- Treats expired provider logins as a human or environment decision, not something to fix in code.
- Helps design test proof for start, resume and restore scenarios.
Install the rig-lifecycle skill
Download the ZIP from the skill's page on this site and unzip it. The folder contains SKILL.md. Then place it in one of two locations: ~/.claude/skills/ to make it available across your projects, or a project's .claude/skills/ folder to limit it to that project.
If you use the Claude apps instead, open Settings, go to Capabilities, then Skills, and upload the skill there.
- Personal install: unzip into ~/.claude/skills/
- Project install: unzip into .claude/skills/ inside the project
- Claude apps: Settings → Capabilities → Skills, then upload
Check the requirements before you start
The skill refers to the rig CLI commands and the OpenRig daemon, so OpenRig must already be installed. Without it the agent has nothing to operate on. The skill does not run scripts and does not use the network. It is guidance text only.
The skill does not perform recovery itself. It shapes how the agent reasons and reports. The actual restore work is done by OpenRig.
- Required: OpenRig installed, with the rig CLI and daemon available
- Runs scripts: no
- Uses network: no
How to trigger the skill with natural requests
Skills load based on their description, so you usually do not need to name the skill. Describe the lifecycle task in plain language and mention the rig or the situation. The skill's description covers requests about lifecycle operations, reading rig ps after recovery, and designing proof for a lifecycle scenario.
Be specific about the situation, such as a reboot, a crash, or a missing provider login. That lets the agent pick the right part of the guidance.
- "My rig came back after a reboot. Can I trust what rig ps shows?"
- "Resume failed for one seat. Walk me through reporting it honestly."
- "Design a test plan for resuming a rig after a host reboot."
- "Check whether this snapshot restore claim is backed by real reboot evidence."
Understand the restore-honesty rule and outcome labels
The central rule is that a failed resume is FAILED loudly, with no automatic fresh fallback. The skill lists restore outcomes so the agent uses the same words you will see in output: resumed means the native runtime resumed the same conversation; rebuilt means a new process was assembled from artifacts; fresh means a new process with no prior continuity; failed means the restore was attempted and failed; attention_required means a recoverable blocker needs operator action; n-a means not applicable.
The skill also separates claims. It says fully_restored is a Rung 3 execution rollup, and that "fully back" is reserved for Rung 4 only. Do not treat these as the same statement.
- resumed, rebuilt, fresh, failed, attention_required, n-a
- Provider auth refusal on Codex returns attention_required; a Claude login prompt returns failed with login_required.
- Cross-runtime alignment is listed as an open follow-up question, so do not assume the two runtimes behave identically.
Use it to design proof for lifecycle scenarios
The skill gives a minimum scenario matrix for proof: clean start, warm resume, host reboot or lost tmux socket, provider auth loss, partial boot or partial failure, and intentional operator recovery from a snapshot.
It separates two tiers. Tier 1 is in-process daemon-unit evidence, which the skill calls bedrock and not sufficient on its own. Tier 2 is a real reboot or a disposable Tart VM. Ask the agent which tier your evidence reaches before you accept a claim of lifecycle proof.
- Clean start: boot from spec into a known state
- Warm resume: rig down, then rig up <name> resumes seats
- Host reboot or tmux socket absence
- Provider auth loss
- Partial boot or partial failure, reported honestly
- Intentional operator recovery from snapshot
rig down
rig up <name>What people use it for
- Recovering a set of agent seats after a laptop reboot without starting from a cleanup job.
- Writing a test plan that covers clean start, warm resume and host reboot for an OpenRig topology.
- Auditing whether a rig's reported health matches what is running in the tmux sessions.
- Deciding whether an expired Claude or Codex login is an operator task rather than a code change.
- Reviewing a restore report to check that a failed resume was not silently replaced by a fresh launch.
Frequently asked questions
How do I install the rig-lifecycle agent skill?
Download the ZIP from this page and unzip it. Put the folder in ~/.claude/skills/ for personal use or in .claude/skills/ inside a project. In the Claude apps, you can instead upload it under Settings → Capabilities → Skills.
Do I need OpenRig installed to use this skill?
Yes. The skill refers to rig CLI commands and the OpenRig daemon, so OpenRig must be installed for the skill to be useful.
Is the rig-lifecycle skill free?
The skill is licensed under Apache-2.0. This page does not list a price, so check the source repository for any terms beyond the licence.
Will this skill restore my rig automatically?
No. The skill is guidance text and does not perform recovery itself. It tells the agent how to reason about restores and how to report failures honestly.
What is the difference between rig-lifecycle and openrig-user?
The rig-lifecycle skill covers reasoning about lifecycle operations and recovery. The openrig-user skill covers the CLI surface, so for a single simple command the openrig-user skill is the better fit.
Why does the skill say a failed resume should not fall back to a fresh start?
A silent fresh launch hides that the previous conversation was not recovered. The skill requires failures to be reported as FAILED, and a fresh launch only happens as an explicit follow-up that a person chooses.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
This skill suits developers and operators who run OpenRig rigs and need to recover work after reboots or crashes, check that reported health matches reality, or design proof for lifecycle scenarios. It is guidance rather than automation, so it will not restore anything by itself. If you only need a simple CLI command, or you do not use OpenRig, a general CLI reference or the openrig-user skill is a better choice.
Download the skill