Skip to content

mvschwarz/openrig / packages/daemon/assets/skills/rigs

Use when a person in their own Claude Code, Codex or OMP (Oh My Pi) session mentions OpenRig, rigs or the rig command, or wants a team of coding agents (Claude Code, Codex) on their project. That includes installing, setting up, starting or updating OpenRig, seeing, getting back to or re-attaching to their OpenRig agents (including bringing a stopped team back), TUI or operator, starting or sharing a team, and joining this session to one. Operating a running rig from inside it (fleet recovery, seat handover, watchdog wakes, in-rig upgrades) is the openrig-skills skill.

FileKindSize
SKILL.mdskill md43.4 KB
references/installing-for-someone.mdmarkdown5.3 KB

references/installing-for-someone.md

View on GitHub
# Installing OpenRig for someone

The agent that installed OpenRig for a person follows step 2 of the `rigs` skill. This note adds what that step does not spell out
(starting the daemon and kernel, what counts as ready and as finished) and relays the same commands; if the two ever differ, step 2 is
right. A seat's startup text carries only a pointer to it.

After setup and a working selected login, use `rig daemon start` if stopped. On a fresh instance it also starts the
kernel, including the operator and advisor. They are part of installation; choosing a project team comes later.
Keep the kernel for a normal install. `rig daemon start --no-kernel` is for automation or an explicit request to
omit its startup, including the operator that helps the person start a team.
Read `rig status` and `rig ps --nodes --rig kernel`: started is not ready. The view may open
while agents finish starting; report their actual state and use the guide's existing recovery routes when needed.
Right after OpenRig is installed or started, and whenever the person wants to see their agents or OpenRig again,
open the OpenRig TUI and operator for them: say so, then run `rig terminal open saved:kernel --window --json` on the
daemon's desktop. Don't wait for particular words. The command opens a new terminal tab/window using
installed herdr, otherwise plain tmux; inside Herdr it opens a space there. It preserves existing conversations. The default view
needs no YAML edit. Below 120 measured columns (or unknown width), show the operator alone first; from 120,
show dashboard and operator equally side by side. The advisor always has a separate tab or tmux window.
This applies to Claude-only, Codex-only and mixed kernels.
Keep the queue worker off the first view and accessible through the TUI. Reuse existing conversations and accounts.
Inspect the result and visible content using authorized desktop tools, or report that visibility is unconfirmed.
Herdr is visible only inside a terminal the person can see. Creating a workspace or switching the shared TUI
to `:terminals` does not open that terminal. A window request or CLI success alone is not visual proof. On a
desktop, do not finish by showing a table or suggesting a command instead of opening the requested view.
Only if the window cannot open, `rig tui --shared` is the named dashboard-only fallback, not the operator's
conversation. Explain the failure and help with the chosen fallback; do not attach in your own terminal.
Over headless SSH, explain that no visible terminal opened; ask the person to open a new terminal window or tab on the
daemon's host (over SSH, a new SSH session to it) and relay the command OpenRig printed, exactly, rather than composing one.
No, SSH and headless use are valid background outcomes. If OpenRig printed no command, find the `operator.agent` row with
`rig ps --nodes --rig kernel --json --fields logicalId,canonicalSessionName,tmuxAttachCommand`. Give `env -u TMUX` followed by
that row's `tmuxAttachCommand`, exactly as listed (it names the daemon's tmux server when that isn't the default
one), for a new terminal on the same host and user (over SSH, connect there first). Ctrl-b, then d detaches without stopping it.
Ask the person's goal, project folder and branch unless they've said, then hand them to the ready operator; do not implement the project yourself.
Find the `operator.agent` row with `rig ps --nodes --rig kernel --json` and use its `canonicalSessionName`. By default the
person stays with you: hand the goal over as a queue task with a return route, so most of the team's questions come back
to you (choice dialogs and other hosts keep their path), and the result is in the handed-over task's resolution. Write
the task as the skill's step 2 says ("If the person stays with you"):
`rig queue create --destination <canonicalSessionName> --return-route auto --summary "<the goal in a sentence>" --body-file <file> --verify`
(if this session joined, `--return-route self` with your `OPENRIG_SESSION_NAME` in front). Only if the person chose to answer
in the operator's pane, send
`rig send <canonicalSessionName> 'This is the agent that installed OpenRig. The person will answer in your pane. Goal: <goal>. Project folder: <absolute path>. Branch: <branch>.'`
instead, or have them type the goal and folder there. Show where the operator answers and let it
help them choose a team before any starter launch.
If the person gives you a goal later, forward it, the folder and constraints the same way, by the route they chose;
leave implementation with the operator's team instead of taking over the work yourself.
Installation is complete when the operator is ready and the person is talking to it or, when they stay with you, once
the goal has reached it as a task with your return route, not merely when the daemon is healthy. If they choose to talk later, leave the exact connection step and say that the handoff is still pending.

Read `rig context get reference/getting-started.md#open-the-kernel-conversations` for the commands and manual
path. Without a working daemon, the same guide is `daemon/docs/reference/getting-started.md` in the installed
`@openrig/cli` package. Confirm what actually opened; on a headless host or unavailable display, say so and give the
person the exact connection/attach step. Opening a terminal is not permission to start or reset seats or provision accounts.
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall