OpenRig

The terminal UI

Documented against OpenRig 0.5.14. Help text uses "node" where these pages say seat, for a seat's position in the running rig.

What it is for

rig tui opens mission control: the one screen where you see the whole fleet and the work it is doing without being the router between seats. The left side is an explorer over what exists: hosts, rigs, pods, seats, specs, projects, missions and slices. The right side shows the selected thing on tabs: a table, recent events, an overview, a graph, health, topology, YAML. A command bar at the top takes the same commands the table below lists, and ? opens a palette that fuzzy-finds every one of them. It is the surface built for a person; the CLI and the terminals are the surface built for agents. Both read the same records.

The tour on the homepage is a staged version of this screen with fictional data. The real one is what you get after rig up.

The three commands you will use first

Open it. Standalone by default; --shared joins the kernel's existing shared terminal and you detach with Ctrl-b then d.

rig tui
rig tui --shared

Find your way. Inside the TUI, type a rig, pod or seat name to drill into it, back to return, and tab to switch the content pane between table, recent, overview, graph, health, topology, configuration, yaml and pulse.

rig my-rig
pod dev
agent impl
tab graph
back

See the work. projects chooses a project from the workspace catalog, mission <name> opens its missions and workflows, source reads the selected project, mission or slice file, and attention (also needs or feed) shows human requests separately from outcome and health updates.

projects
mission search-experience
source
attention

Every TUI command

This table is the TUI's own command registry, listed by rig tui commands (add --json for agents).

Command Aliases Arguments What it does
terminals browse Saved and Derived terminal views; preview before explicit Open
terminal-preview <view> passively preview a saved:id or rig:name terminal view
attention needs, feed inspect human requests and outcome/health updates
read <root>/<path>[#heading] read a current file within an explicitly configured root
system instance Health, Configuration and Connections
config [category] browse instance settings; Slack is one category
setting <key> open a setting with its full value, source and scope
refresh read the current view again; stored values do not prove runtime adoption
timezone show local time setting and persistent rig config instructions
recent <transition-id> inspect an original event from the served Recent window
connections System Connections: gateway, recipients and routes
back return to the previous view, selection and scroll
projects choose a project from the workspace catalog
project <id> select an exact catalog project
source read the selected project, mission or slice source
mission <name> open a mission's work and workflows
workflow <instance-id> open a workflow in the selected mission
packet <qitem-id> open current workflow work in the selected mission
: <section> jump to a section
/ <text> filter rows by text
tab <table / recent / overview / graph / health / topology / configuration / yaml / pulse> switch the content-pane view tab
graph g open the topology graph view
style <name> set the graph render style (validated by dispatch against the style registry)
scroll <up / down> scroll the content pane
select-text copy toggle terminal-native drag selection and copy
top scroll the content pane to the top
bottom scroll the content pane to the bottom
find <text> filter rows by text (verb form of the / prefix)
spec-of <agent> cross-navigate to the spec of the named agent
running <spec> cross-navigate to agents running the named spec
help ? open the command palette (fuzzy-find every command)
reqs toggle mini-requirements collapse (scopes view)
narrative toggle the PROGRESS.md narrative panel (scopes view)
host <name> drill into the named host
rig <name> drill into the named rig
pod <name> drill into the named pod
agent <name> drill into the named agent
spec <name> drill into the named spec

What it does not do

  • It does not start agents, launch terminals or resume seats by itself. terminals previews a saved or derived view; opening tiles is an explicit action that needs a terminal provider (rig terminal open).
  • It does not replace the CLI for agents. Every command an agent needs is a rig verb with --json.
  • The older web UI (rig ui open) still exists as a verb but is unmaintained and replaced by the TUI. Do not diagnose product behaviour from it.
  • Reading a request in attention does not approve or deliver it. Resolved settings shown under system do not prove runtime adoption.

Where it goes next

  • Messaging: how seats reach each other and how you reach one seat's screen.
  • Coordination: the queue rows the attention and pickup views are built from.
  • Configuration: the keys the system view browses, with their defaults and sources.